# Import modules
import os

import pandas as pd
import numpy as np
import scipy.stats as stats

import re # regex
import missingno as msno # matrix for missing values

# Plotting.
import seaborn as sns
import matplotlib.pyplot as plt
import squarify
%matplotlib inline

from datetime import datetime

import warnings
warnings.filterwarnings('ignore')

pd.set_option('future.no_silent_downcasting', True)

# Load the HR dataset into a dataframe.
df = pd.read_csv("HRDataset_v14.csv", header=0, sep=',')

# Check for inifinite values np.inf.
numeric_df = df._get_numeric_data()
inf_count = np.isinf(numeric_df).values.sum()
print("Contains " + str(inf_count) + " infinite values") 

# Check for duplicate rows.
duplicate_rows = df.duplicated().sum()
print("Contains " + str(duplicate_rows) + " duplicate rows")

Contains 0 infinite values
Contains 0 duplicate rows

# Check for values that contain question mark.
print(f"Count of values with ?: {df.eq('?').sum().sum()}")

Count of values with ?: 0

# Function to check for leading and trailing spaces in an object column.
def check_leading_trailing_spaces(column):
    return column.str.strip().fillna('') != column.fillna('')

# Function to store columns with leading or trailing spaces in a Dictionary
def check_columns_for_spaces(df):
    columns_with_spaces = {}  # Dictionary to store columns.
    for col in df.select_dtypes(include='object').columns:
        has_spaces = check_leading_trailing_spaces(df[col])
        if has_spaces.any():  # Check if any leading or trailing spaces are found
            columns_with_spaces[col] = has_spaces.sum()  # Store the count of rows with spaces
    return columns_with_spaces

# Function to Trim columns with leading or trailing spaces.
def trim_columns(df, cols):
    for col in cols:
        df[col] = df[col].str.strip()

# Call the function to check for spaces in all object type columns.
columns_with_spaces = check_columns_for_spaces(df)

print("Columns with leading or trailing spaces and their counts:")
for col, count in columns_with_spaces.items():
    print(f"{col}: {count}")

Columns with leading or trailing spaces and their counts:
Employee_Name: 70
Position: 1
Sex: 135
Department: 209

# Iterate over each column name with leading or trailing spaces.
for col, count in columns_with_spaces.items():
    # Check for leading or trailing spaces in the current column.
    has_spaces = check_leading_trailing_spaces(df[col])
    # Filter rows with leading or trailing spaces in the current column.
    rows_with_spaces = df[has_spaces]
    
    # Print column name.
    print(f"Sample of Rows with leading or trailing spaces in '{col}' column:")
    if not rows_with_spaces.empty:
        print(rows_with_spaces[col].head(3))
    else:
        print("No rows with leading or trailing spaces found in this column.")
    print()  # Add an empty line for separation between columns.

Sample of Rows with leading or trailing spaces in 'Employee_Name' column:
1    Ait Sidi, Karthikeyan   
4            Anderson, Carol 
5           Anderson, Linda  
Name: Employee_Name, dtype: object

Sample of Rows with leading or trailing spaces in 'Position' column:
249    Data Analyst 
Name: Position, dtype: object

Sample of Rows with leading or trailing spaces in 'Sex' column:
0    M 
1    M 
7    M 
Name: Sex, dtype: object

Sample of Rows with leading or trailing spaces in 'Department' column:
0    Production       
2    Production       
3    Production       
Name: Department, dtype: object

# Call the function to Trim columns.
trim_columns(df, columns_with_spaces)

# Call the function to check for spaces again.
columns_with_spaces = check_columns_for_spaces(df)

# Verification - Check if any leading or trailing spaces remain after trimming.
verification_result = not columns_with_spaces
print("Verification - Columns contain no leading or trailing spaces:", verification_result)

Verification - Columns contain no leading or trailing spaces: True

# Filter rows where 'DOB' has length 8.
dob_2_digit_years = df[df['DOB'].str.len() == 8]

# Select a random instance from the filtered Dataframe.
random_test_instance = dob_2_digit_years.sample(random_state=42) # Use a fixed random state for reproducibility.

# Test instance.
random_test_instance_index = random_test_instance.index[0]
dob_value = random_test_instance['DOB'].iloc[0]
print(f'DOB value on test instance (index={random_test_instance_index}):', dob_value)

DOB value on test instance (index=290): 10/10/68

dob_test = df.copy()

dob_test["DOB"] = pd.to_datetime(dob_test['DOB'], format='%m/%d/%y', errors='coerce')
dob_test['DOB'].to_csv('dob.csv', index=False, header=['DOB'])
print("Inspect dob.csv export for confirmation of issue.\n")

# Extract year component and find min max of Year.
dob_test['Year'] = dob_test['DOB'].dt.year
year_min = dob_test['Year'].min()
year_max = dob_test['Year'].max()
print(f"Range of years: {year_min} to {year_max}\n")

# Calculate age for each employee with this issue.
current_year = pd.Timestamp.now().year
dob_test['Age'] = current_year - dob_test['Year']
print(f"Average age of employee with issue: {dob_test['Age'].mean():.0f}\n")

Inspect dob.csv export for confirmation of issue.

Range of years: 1969 to 2068

Average age of employee with issue: 32

# Define a function to parse dates.
def parse_date(date_str):

    # This retrieves all characters of the string after the last forward slash.
    year_str = date_str.split('/')[-1]

    # If the year is only 2 digits, append '19' or '20' to it 
    if len(year_str) == 2:            
        
        current_year = datetime.now().year
        if int(year_str) <= current_year % 100:         
            # Append '20' if the year is less than or equal to the last two digits of the current year (24) 
            # e.g. DOB is 2006, assuming person is 18 years old not 118.
            date_str = date_str[:-2] + '20' + date_str[-2:]
        else:
            # Append the string '19' to represent the 20th century.
            # This retrieves the date up until last two digits, appends 19 then retrieves last 
            # two characters of the string, which represent the two-digit year
            date_str = date_str[:-2] + '19' + date_str[-2:]            
        
    try:
        # Attempt to parse the date assuming a month/day/year format
        return datetime.strptime(date_str, '%m/%d/%Y').date()
    except ValueError:
        pass
    try:
        # Attempt to parse the date with a different format
        return datetime.strptime(date_str, '%m/%d/%y').date()
    except ValueError:
        pass
    
    # If parsing fails, return None
    return None

# Copy original dataframe for testing purposes.
parse_date_df_test = df.copy()

# Apply the parse_date function to the 'DOB' column
parse_date_df_test['Parsed_DOB'] = parse_date_df_test['DOB'].apply(parse_date)

# Extract year, month, and day into separate columns
parse_date_df_test[['Year', 'Month', 'Day']] = pd.DataFrame([(d.year, d.month, d.day) if d else (None, None, None) for d in parse_date_df_test['Parsed_DOB']])

# Test function to print range of years and average age.
def test_dob_parsing(df):
    
    # Find min max of Year
    year_min = parse_date_df_test['Year'].min()
    year_max = parse_date_df_test['Year'].max()
    print(f"Range of years: {year_min} to {year_max}\n")

    # Calculate average age for employees.
    current_year = pd.Timestamp.now().year
    parse_date_df_test['Age'] = current_year - parse_date_df_test['Year']
    print(f"Average age of employee with issue rectified: {parse_date_df_test['Age'].mean():.0f}\n")

# Confirm parsing success.
test_dob_parsing(parse_date_df_test)

# Display the test case.
parsed_dob_value = parse_date_df_test.loc[random_test_instance_index, 'Parsed_DOB'].strftime('%m/%d/%Y')
print("Parsed_DOB value on test instance:", parsed_dob_value)

Range of years: 1951 to 1992

Average age of employee with issue rectified: 45

Parsed_DOB value on test instance: 10/10/1968

# Apply the parse_date function to the 'DOB' column
df['Parsed_DOB'] = df['DOB'].apply(parse_date)

# Extract year, month, and day into separate columns
df[['Year', 'Month', 'Day']] = pd.DataFrame([(d.year, d.month, d.day) if d else (None, None, None) for d in df['Parsed_DOB']])

# Display the dataframe.
df.head()

# Confirm issue addressed on original dataframe.
test_dob_parsing(df)

# Confirm test instance applied.
parsed_dob_value = df.loc[random_test_instance_index, 'Parsed_DOB'].strftime('%m/%d/%Y')
print("Parsed_DOB value on test instance:", parsed_dob_value)

Range of years: 1951 to 1992

Average age of employee with issue rectified: 45

Parsed_DOB value on test instance: 10/10/1968

# Calculate age for each employee.
current_year = pd.Timestamp.now().year
df['Age'] = current_year - df['Year']
df[['Age', 'Parsed_DOB']].sample()

df.head(3)

df.sample(3)

# Inspect non-numeric columns.
df.select_dtypes(include=['object']).head(3)

# Print instance and column counts.
print(f"HR dataset contains: \nRows: \t\t{df.shape[0]}")
print(f"Columns: \t{df.shape[1]}")

HR dataset contains: 
Rows: 		311
Columns: 	41

# List the data types of each attribute.
print(df.dtypes.sort_values())

Age                             int64
EmpSatisfaction                 int64
Day                             int64
SpecialProjectsCount            int64
DaysLateLast30                  int64
Absences                        int64
Zip                             int64
Year                            int64
PositionID                      int64
Termd                           int64
Month                           int64
FromDiversityJobFairID          int64
PerfScoreID                     int64
DeptID                          int64
EmpStatusID                     int64
GenderID                        int64
MaritalStatusID                 int64
MarriedID                       int64
EmpID                           int64
Salary                          int64
EngagementSurvey              float64
ManagerID                     float64
RecruitmentSource              object
Parsed_DOB                     object
LastPerformanceReview_Date     object
PerformanceScore               object
Employee_Name                  object
Department                     object
EmploymentStatus               object
TermReason                     object
DateofTermination              object
DateofHire                     object
HispanicLatino                 object
CitizenDesc                    object
MaritalDesc                    object
Sex                            object
DOB                            object
State                          object
Position                       object
ManagerName                    object
RaceDesc                       object
dtype: object

missing_counts = df.isna().sum()

# Display only attributes with missing values.
missing_counts_filtered = missing_counts[missing_counts > 0]
print(f"Missing values:\n{missing_counts_filtered}")

Missing values:
DateofTermination    207
ManagerID              8
dtype: int64

# Missingno matrix - positional information of the missing values.
msno.matrix(df[['DateofTermination', 'ManagerID']], figsize=(3,4), fontsize=10,
            color=(135/255, 185/255, 210/255)) # Tuple of RGB colors.

<Axes: >

# Calculate Attrition Rate
total_employees = df.shape[0]
print(f"Total Employees: {total_employees}")

total_terminated = df[df['EmploymentStatus'] != 'Active'].shape[0]
print(f"Total Terminated: {total_terminated}\n")

attrition_rate = (total_terminated / total_employees) * 100
print(f"ATTRITION RATE: {attrition_rate:.2f}%")

Total Employees: 311
Total Terminated: 104

ATTRITION RATE: 33.44%

# Summary statistics of all numerical attributes.
df.describe(include='number')

# Exclude the column names containing ID.
columns_to_exclude = [col for col in df.columns if 'ID' in col]

# Drop the columns.
filtered_df = df.drop(columns=columns_to_exclude)

# Use describe() on the filtered dataframe.
describe_no_id = filtered_df.describe(include='number')

# Display summary statistics.
describe_no_id

# Summary statistics of all non-numerical attributes.
df.describe(include='object')

# Calculate unique counts for specific ID columns to explore the distribution of categorical variables.
columns_of_interest = ['GenderID', 'MaritalStatusID', 'MarriedID', 'EmpStatusID']
for col in columns_of_interest:
    unique_counts = df[col].value_counts()
    print(f"Unique counts for column '{col}':\n{unique_counts}\n")

Unique counts for column 'GenderID':
GenderID
0    176
1    135
Name: count, dtype: int64

Unique counts for column 'MaritalStatusID':
MaritalStatusID
0    137
1    124
2     30
3     12
4      8
Name: count, dtype: int64

Unique counts for column 'MarriedID':
MarriedID
0    187
1    124
Name: count, dtype: int64

Unique counts for column 'EmpStatusID':
EmpStatusID
1    184
5     88
3     14
4     14
2     11
Name: count, dtype: int64

# Histograms of continuous variables.
ax = df.hist(bins=25, figsize=(12,8),                
                edgecolor='blue', 
                facecolor='lightblue',
                grid=True)

# Set facecolor.
for axes in ax.flatten():
    axes.set_facecolor('#F0F0F0')
    
plt.tight_layout(pad=1.5)
plt.show()

# Investigate distribution and any outliers on specific continuous variables.
scatterplot_cols = ['Salary', 'EngagementSurvey', 'Absences', 'SpecialProjectsCount']

fig, ax = plt.subplots(1,len(scatterplot_cols), figsize=(10,4)) 

# Set facecolor.
for axes in ax.flatten():
    axes.set_facecolor('#F0F0F0')

for idx, col in enumerate(df[scatterplot_cols]):
    sns.scatterplot(data = df[col], ax = ax[idx], color='lightblue', edgecolor='blue')
    ax[idx].set_ylabel(col, fontweight='bold')  # Set y-axis label to bold
    
fig.tight_layout()
fig.show()

# Calculate description statistics
description = df[['Salary', 'EngagementSurvey']].describe()

# Calculate IQR for Salary
Q1_salary = description.loc['25%', 'Salary']
Q3_salary = description.loc['75%', 'Salary']
IQR_salary = Q3_salary - Q1_salary

# Calculate IQR for EngagementSurvey
Q1_engagement = description.loc['25%', 'EngagementSurvey']
Q3_engagement = description.loc['75%', 'EngagementSurvey']
IQR_engagement = Q3_engagement - Q1_engagement

# Identify potential outliers for Salary.
outlier_threshold = 2.5 #1.5 
potential_outliers_salary = (df['Salary'] < (Q1_salary - outlier_threshold * IQR_salary)) | (df['Salary'] > (Q3_salary + outlier_threshold * IQR_salary))

# Identify potential outliers for EngagementSurvey.
potential_outliers_engagement = (df['EngagementSurvey'] < (Q1_engagement - outlier_threshold * IQR_engagement)) | (df['EngagementSurvey'] > (Q3_engagement + outlier_threshold * IQR_engagement))

# Threshold adjustment - determined after visual analysis of scatterplots.
print(f"Adjusted outlier threshold to: {outlier_threshold}\n")

# Print potential outliers for Salary.
print("Potential outliers for Salary:")
print(df[potential_outliers_salary]['Salary'].sort_values(ascending=False))

# Print potential outliers for EngagementSurvey.
print("\nPotential outliers for EngagementSurvey:")
print(df[potential_outliers_engagement]['EngagementSurvey'].sort_values(ascending=False))

Adjusted outlier threshold to: 2.5

Potential outliers for Salary:
150    250000
308    220450
131    180000
96     178000
55     170500
190    157000
240    150290
244    148999
243    140920
76     138888
239    120000
108    114800
292    113999
Name: Salary, dtype: int64

Potential outliers for EngagementSurvey:
69    1.12
Name: EngagementSurvey, dtype: float64

# Correlation with Termination status.
termd_corr = df.corr(numeric_only=True)['Termd'].sort_values()
termd_corr.corr

<bound method Series.corr of SpecialProjectsCount     -0.147429
Zip                      -0.139006
Year                     -0.101995
Salary                   -0.093994
PerfScoreID              -0.089061
EngagementSurvey         -0.017543
GenderID                 -0.015741
Day                      -0.015623
EmpSatisfaction          -0.004732
Month                     0.005216
DeptID                    0.065922
MarriedID                 0.077028
EmpID                     0.092389
Absences                  0.098274
MaritalStatusID           0.099367
Age                       0.101995
DaysLateLast30            0.136379
PositionID                0.147042
FromDiversityJobFairID    0.147717
ManagerID                 0.209113
EmpStatusID               0.948058
Termd                     1.000000
Name: Termd, dtype: float64>

# Plot heatmap for correlation analysis.
corr_matrix = df.corr(numeric_only=True)
plt.figure(figsize=(10, 8))
sns.heatmap(corr_matrix, annot=True, cmap='magma', fmt=".2f", annot_kws={"size": 6})
plt.xticks(fontsize=7)
plt.yticks(fontsize=7)
plt.title('Correlation Matrix for Numeric Columns')
plt.show()

# Function to generate pastel colours for visuals.
def generate_pastel_palette(num_colors):
    return sns.color_palette("pastel", num_colors, desat=0.95)

# Configure settings and colors for charts.
chart_bg_color = '#FFFFF5' # Pale yellow.
bar_edge_color = 'darkblue'
pie_wedge_color = 'darkblue'
annotation_color = 'darkblue'
scatter_edge_color = 'darkblue'

# Define common label parameter values in a dictionary.
title_label_params = {'fontweight': 'bold', 'pad': 10}

# Count the occurrences of each reason for termination.
employment_status_counts = df['EmploymentStatus'].value_counts()

# Generate a pastel color palette based on the number of termination reasons.
pastel_colors = generate_pastel_palette(len(employment_status_counts))

# Create a list of explode values to increase space between slices.
explode = [0.028] * len(employment_status_counts)

plt.figure(figsize=(5, 5))
patches, texts, _ = plt.pie(employment_status_counts, 
        labels=employment_status_counts.index, 
        autopct='%1.1f%%',
        startangle=128,
        colors=pastel_colors,
        explode=explode,
        shadow=True,
        textprops={'color': annotation_color, 'weight': 'normal', 'fontsize': '9'},
        wedgeprops={'linewidth': 0.3, 'edgecolor': pie_wedge_color},
        labeldistance=1.09)
plt.axis('equal')  # Equal aspect ratio ensures that pie is drawn as a circle.
plt.title('Employment Status', **title_label_params)

# Set font color of text labels.
for text in texts:
    text.set_color('black')
    text.set_fontsize('9')
    
plt.show()

# Count the occurrences of each reason for termination.
reason_counts = df[df['TermReason'] != 'N/A-StillEmployed']['TermReason'].value_counts()

# Generate a pastel color palette based on the number of termination reasons.
pastel_colors = generate_pastel_palette(len(reason_counts))

# Create a horizontal bar plot.
fig, ax = plt.subplots(figsize=(8, 6))
reason_counts.plot(kind='barh', color=pastel_colors, edgecolor=bar_edge_color, linewidth=0.5)
ax.set_facecolor(chart_bg_color)

# Annotate each bar with its count.
for i, count in enumerate(reason_counts):
    ax.text(count + 0.1, i, str(count), va='center', color=annotation_color)
    
# Adjust the limits of the y-axis to provide more room.
ax.set_ylim(-0.7, len(reason_counts) -  0.2)

 # Hide spines.
ax.spines['top'].set_visible(False) 
ax.spines['right'].set_visible(False) 

# Convert x-labels to have uppercase for first letter.
ax.set_yticklabels([label.capitalize() for label in reason_counts.index])

plt.xlabel('Count', fontweight='bold')
plt.ylabel('Reason for Termination', fontweight='bold')
plt.title('Distribution of Termination Reasons', **title_label_params)
plt.gca().invert_yaxis()  # Invert y-axis to display the most frequent reasons at the top.
plt.grid(True, alpha=0.15)
plt.show()

# Count number of employees per department.
dept_counts = df['Department'].str.strip().value_counts()

# Generate a pastel color palette based on the department.
pastel_colors = generate_pastel_palette(len(dept_counts))

# Plot the bar plot
fig, ax = plt.subplots(figsize=(8,6))
dept_counts.plot(kind='bar', color=pastel_colors, edgecolor=bar_edge_color, linewidth=0.5)
ax.set_facecolor(chart_bg_color)

# Adjust the limits of the y-axis.
ax.set_ylim(0,)

# Hide spines.
ax.spines['top'].set_visible(False) 
ax.spines['right'].set_visible(False)

# Annotate bars with figure just below the top of each bar.
for i, count in enumerate(dept_counts):
    ax.text(i, count + 0.2, str(count), ha='center', va='bottom', color=annotation_color)

plt.xlabel('')
plt.ylabel('Count of Employees', fontweight='bold')
plt.title('Number of Employees per Department', **title_label_params)
plt.xticks(rotation=45, ha='right')
plt.tight_layout()
plt.grid(True, alpha=0.25, axis='y')
plt.show()

# Filter dataframe for columns containing words Gender or Sex. 
gender_df = df.copy()
gender_df = df.loc[:, df.columns.str.contains('Gender', case=False) | df.columns.str.contains('Sex', case=False)]
gender_df.sample(4)

# Drop GenderID column.
gender_df = gender_df.drop('GenderID', axis=1)

# Remove leading and trailing whitespaces from 'Sex' column.
gender_df['Sex'] = gender_df['Sex'].str.strip()

# Gender data.
gender_counts = gender_df.value_counts()
sizes = gender_counts.values
labels = ['Female', 'Male']

# Generate a pastel color palette based on the number of genders.
pastel_colors = generate_pastel_palette(len(gender_counts))

# Get counts of females and males.
female_count = sizes[0] if 'Female' in labels else 0
male_count = sizes[1] if 'Male' in labels else 0

# Create legend labels with additional text.
legend_labels = [f'{label} ({count})' for label, count in zip(labels, sizes)]

# Explosion of pie wedges.
explode = (0.05, 0.05)

# Pie Chart.
fig, ax = plt.subplots(figsize=(6,6))
wedges, texts, autotexts = plt.pie(sizes, 
                                   colors=pastel_colors, 
                                   labels=labels, 
                                   autopct='%1.1f%%', 
                                   pctdistance=0.80, 
                                   explode=explode,
                                   labeldistance=0.35, 
                                   wedgeprops={'linewidth': 0.3, 'edgecolor': pie_wedge_color},
                                   textprops={'color': annotation_color, 'weight': 'normal', 'fontsize': '12'},
                                   shadow=True)

# Set autopct label color to white if the label is 'Male'.
#for autotext, label in zip(autotexts, labels):
#    if label == 'Male':
#        autotext.set_color('white')

# Draw inner circle for donut effect.
centre_circle = plt.Circle((0, 0), 0.65, fc='white')
fig = plt.gcf()

# Adding circle in pie chart.
fig.gca().add_artist(centre_circle)

# Adding title of the chart.
plt.title('Distribution of Gender', **title_label_params)

# Add Legends.
plt.legend(legend_labels, loc="upper right", bbox_to_anchor=(1.04, 1))

# Set font color of text labels.
for text in texts:
    text.set_color('black')
    text.set_fontsize('12')

# Display the chart.
plt.show()

df['RaceDesc'].value_counts()

RaceDesc
White                               187
Black or African American            80
Asian                                29
Two or more races                    11
American Indian or Alaska Native      3
Hispanic                              1
Name: count, dtype: int64

# Count number of employees in each race category.
race_counts = df['RaceDesc'].value_counts().sort_values()

# Generate a pastel color palette.
pastel_colors = generate_pastel_palette(6)

# Plotting
plt.figure(figsize=(8, 6))
ax = race_counts.plot(kind='line', 
                      marker='o', 
                      markersize=8,
                      color=pastel_colors[2], 
                      linewidth=3.0)

# Set the background color of the axis
ax.set_facecolor(chart_bg_color)

# Get the current tick positions and labels.
ticks = ax.get_xticklabels()

# Adjust the position of the tick labels slightly to the left
for tick in ticks:
    tick.set_horizontalalignment('left')
    tick.set_x(-3.75)
    
# Customize x-tick labels
tick_labels = [label.replace(' ', '\n') for label in race_counts.index]  # Replace space with newline character
# Set custom tick labels
plt.xticks(range(len(race_counts)), tick_labels, rotation=0)
  
# Add counters on markers.
for i, count in enumerate(race_counts):
    plt.text(i, count+4, str(count), ha='center', va='bottom', fontsize=10, color=annotation_color)  # Annotate each marker with its count value

# Add labels and title.
plt.title('Count of Employees by Race', **title_label_params)
plt.xlabel('Race Description', fontweight='bold')
plt.ylabel('Count', fontweight='bold')
plt.xticks(rotation=0)

plt.grid(True, alpha=0.20)
plt.ylim(-10,210)
plt.tight_layout()
plt.show()

# Checking integrity of hispanic count.
hispanic_employees = df[df['RaceDesc'].str.contains('hispanic', case=False)]
hispanic_employees

df['Age'].describe()

count    311.000000
mean      45.408360
std        8.870236
min       32.000000
25%       38.000000
50%       44.000000
75%       51.000000
max       73.000000
Name: Age, dtype: float64

# Set up the figure and axis
plt.figure(figsize=(8, 6))

# Generate a pastel color palette.
pastel_colors = generate_pastel_palette(2)

# Plot the histogram
ax = sns.histplot(df['Age'], kde=True, kde_kws=dict(cut=3), color=pastel_colors[1], edgecolor=bar_edge_color)

# Set background color behind the plot area
ax.set_facecolor(chart_bg_color)

# Add labels and title
plt.title('Distribution of Age', fontweight='bold')
plt.xlabel('Age', fontweight='bold')
plt.ylabel('Frequency',fontweight='bold')

# Add grid lines
plt.grid(True, linestyle='--', alpha=0.15)

# Customize x-axis tick intervals
plt.xticks(range(20, 90, 5))  # Setting intervals every 5 years from [20 to 90)

# Add mean, median, and quartile annotations.
plt.axvline(x=df['Age'].mean(), color='red', linestyle='--', label='Mean: {:.2f}'.format(df['Age'].mean()))
plt.axvline(x=df['Age'].median(), color='blue', linestyle='--', label='Median: {:.2f}'.format(df['Age'].median()))
plt.axvline(x=df['Age'].quantile(0.25), color='green', linestyle='--', label='25th Percentile: {:.2f}'.format(df['Age'].quantile(0.25)))
plt.axvline(x=df['Age'].quantile(0.75), color='purple', linestyle='--', label='75th Percentile: {:.2f}'.format(df['Age'].quantile(0.75)))

# Show legend
plt.legend()

# Show plot
plt.show()

# Count the occurrences of each performance score.
perfscore_counts = df['PerformanceScore'].value_counts()

# Generate a pastel color palette based on the number of termination reasons.
pastel_colors = generate_pastel_palette(len(perfscore_counts))

# Create a list of explode values to increase space between slices.
explode = [0.028] * len(perfscore_counts)

plt.figure(figsize=(5, 5))
patches, texts, _ = plt.pie(perfscore_counts, 
        labels=perfscore_counts.index, 
        autopct='%1.1f%%',
        startangle=188,
        colors=pastel_colors,
        explode=explode,
        shadow=True,
        textprops={'color': annotation_color, 'weight': 'normal', 'fontsize': '9'},
        wedgeprops={'linewidth': 0.3, 'edgecolor': pie_wedge_color},
        labeldistance=1.09)
plt.axis('equal')  # Equal aspect ratio ensures that pie is drawn as a circle.
plt.title('Performance Score Distribution', **title_label_params)

# Set font color of text labels.
for text in texts:
    text.set_color('black')
    text.set_fontsize('9')

# Calculate mean salary for each gender.
mean_salary_by_gender = df.groupby('Sex')['Salary'].agg(['mean', 'median'])
mean_salary_by_gender

# Top 5 salaries for female.
df[df['Sex'] == 'F']['Salary'].sort_values(ascending=False).head(5)

150    250000
308    220450
131    180000
240    150290
239    120000
Name: Salary, dtype: int64

# Filter the dataframe by gender salaries.
female_salaries = df[df['Sex'].str.strip() == 'F']['Salary']
male_salaries = df[df['Sex'].str.strip() == 'M']['Salary']

# Create a figure and axis object.
fig, ax = plt.subplots(figsize=(8,6))

# Generate a pastel color palette with 6 colors to choose 2 from.
pastel_colors = generate_pastel_palette(6)

# Set the background color of the axis.
ax.set_facecolor(chart_bg_color)

# Marker data point size.
marker_size = 60

# Create a scatterplot for female salaries.
plt.scatter(range(len(female_salaries)), female_salaries, 
            color=pastel_colors[2], 
            alpha=0.7,
            edgecolor=scatter_edge_color,
            s=marker_size, # Set marker size
            label='Female')  

# Create a scatterplot for male salaries.
plt.scatter(range(len(male_salaries)), male_salaries, 
            color=pastel_colors[5], 
            alpha=0.5, 
            edgecolor=scatter_edge_color,
            s=marker_size, # Set marker size
            label='Male')  

# Set the title and labels
plt.title('Female and Male Salaries', **title_label_params)
plt.xlabel('Index') # index position in the dataframe
plt.ylabel('Salary', fontweight='bold')
plt.tight_layout()
plt.grid(True, alpha=0.25)

# Format y-tick labels directly
#yticks = ax.get_yticks()
yticks = np.arange(20000, max(df['Salary'])+20000, 20000)

ytick_labels = []  # Initialise an empty list to store the formatted tick labels
for y in yticks:
    # Format the current tick value y as a string with dollar sign, comma separator, and zero decimal places.
    formatted_label = '${:,.0f}'.format(y)
    # Append the formatted label to the ytick_labels list.
    ytick_labels.append(formatted_label)

ax.set_yticks(yticks)
ax.set_yticklabels(ytick_labels)

# Set y-axis limit to start at 30,000.
ax.set_ylim(bottom=30000)

# Show the legend
ax.legend()

# Show the plot
plt.show()

# Dataframes by specific gender and salary column.
female_salaries = df[df['Sex'].str.strip() == 'F']['Salary']
male_salaries = df[df['Sex'].str.strip() == 'M']['Salary']

# Create a figure and set of subplots.
fig, axes = plt.subplots(1, 2, figsize=(12, 6))

# Generate a pastel color palette with 6 colors to choose 2 from.
pastel_colors = generate_pastel_palette(6)

# Calculate the maximum salary for both genders
max_salary = max(female_salaries.max(), male_salaries.max())

# Calculate the maximum y-tick value based on the maximum salary
max_ytick = int(np.ceil(max_salary / 25000)) * 25000

# Create the violin plot for female salaries
sns.violinplot(y=female_salaries, color=pastel_colors[2], ax=axes[0])
axes[0].set_title('Distribution of Female Salaries', **title_label_params)
axes[0].set_ylabel('Salary', fontweight='bold')
axes[0].set_yticks(np.arange(0, max_ytick + 1, 25000))  # Adjust y-axis tick intervals
axes[0].grid(True, alpha=0.25)
axes[0].set_facecolor(chart_bg_color) # Set the background color of the axis

# Create the violin plot for male salaries
sns.violinplot(y=male_salaries, color=pastel_colors[5], ax=axes[1])
axes[1].set_title('Distribution of Male Salaries', **title_label_params)
axes[1].set_ylabel('Salary', fontweight='bold')
axes[1].set_yticks(np.arange(0, max_ytick + 1, 25000))  # Adjust y-axis tick intervals
axes[1].grid(True, alpha=0.25)
axes[1].set_facecolor(chart_bg_color) # Set the background color of the axis

# Format the tick values y as a string with dollar sign, comma separator, and zero decimal places.
yticks = np.arange(0, max_ytick + 1, 25000)
ytick_labels = ['${:,.0f}'.format(y) for y in yticks]
axes[0].set_yticklabels(ytick_labels)
axes[1].set_yticklabels(ytick_labels)

# Adjust layout to add padding between subplots.
plt.subplots_adjust(wspace=0.4)

# Show the plot
plt.show()

# Generate a pastel color palette with 6 colors to choose 2 from.
pastel_colors = generate_pastel_palette(6)

# Concatenate the data for plotting
combined_salaries = pd.concat([female_salaries, male_salaries], axis=1)
combined_salaries.columns = ['Female', 'Male']  # Rename columns for clarity

# Plot figure.
plt.figure(figsize=(6,6))

# Create the box plot for combined salaries
ax = sns.boxplot(data=combined_salaries, palette=[pastel_colors[2], pastel_colors[5]])

ax.set_facecolor(chart_bg_color) # Set chart background colour.

# Format the tick values y as a string with dollar sign, comma separator, and zero decimal places.
yticks = np.arange(0, max_ytick + 1, 25000)
ytick_labels = ['${:,.0f}'.format(y) for y in yticks]

# Set yticks and yticklabels
ax.set_yticks(yticks)
ax.set_yticklabels(ytick_labels)

# Set y-axis limit to start around $30,000
ax.set_ylim(bottom=30000)

# Set the title and labels
plt.title('Distribution of Salaries by Gender', **title_label_params)
plt.ylabel('Salary', fontweight='bold')
plt.xlabel('')
plt.grid(True,alpha=0.25)

# Show the plot
plt.show()

# Copy dataframe and then remove whitespaces from 'Sex' and 'Position' columns.
df_positions = df.copy()
df_positions['Position'] = df_positions['Position'].str.strip()
df_positions['Sex'] = df_positions['Sex'].str.strip()

# Distribution of positions in the company.
df_positions['Position'].value_counts()

Position
Production Technician I         137
Production Technician II         57
Area Sales Manager               27
Production Manager               14
Software Engineer                10
IT Support                        8
Data Analyst                      8
Network Engineer                  5
Sr. Network Engineer              5
Database Administrator            5
BI Developer                      4
Sales Manager                     3
Accountant I                      3
Senior BI Developer               3
Administrative Assistant          3
Sr. Accountant                    2
IT Manager - DB                   2
Sr. DBA                           2
President & CEO                   1
Data Architect                    1
Principal Data Architect          1
IT Manager - Infra                1
Shared Services Manager           1
IT Director                       1
Director of Sales                 1
IT Manager - Support              1
Director of Operations            1
BI Director                       1
Software Engineering Manager      1
Enterprise Architect              1
CIO                               1
Name: count, dtype: int64

# Mean salary for each position.
df_positions.groupby('Position')['Salary'].mean().round(2).sort_values(ascending=False)

Position
President & CEO                 250000.00
CIO                             220450.00
Director of Sales               180000.00
IT Director                     178000.00
Director of Operations          170500.00
IT Manager - Infra              157000.00
Data Architect                  150290.00
IT Manager - DB                 144959.50
IT Manager - Support            138888.00
Principal Data Architect        120000.00
BI Director                     110929.00
Database Administrator          108499.60
Enterprise Architect            103613.00
Sr. Accountant                  102859.00
Sr. DBA                         102234.00
Software Engineer                96719.20
BI Developer                     95465.00
Sr. Network Engineer             93070.80
Shared Services Manager          93046.00
Data Analyst                     89756.88
Senior BI Developer              84802.67
Software Engineering Manager     77692.00
Production Manager               75294.50
Sales Manager                    69240.00
Area Sales Manager               64932.56
Production Technician II         64892.21
IT Support                       63684.38
Accountant I                     63507.67
Network Engineer                 61605.00
Production Technician I          55524.18
Administrative Assistant         52280.00
Name: Salary, dtype: float64

# Define a function to create a pie chart for a specified salary range.
def create_salary_range_pie(ax, df, salary_min, salary_max):
    # Set the dataframe to the specified salary range.
    salary_range_df = df[(df['Salary'] >= salary_min) & (df['Salary'] <= salary_max)]    

    # Count the number of males and females within the specified salary range
    male_count = salary_range_df[salary_range_df['Sex'].str.strip() == 'M'].shape[0]
    female_count = salary_range_df[salary_range_df['Sex'].str.strip() == 'F'].shape[0]
    sizes = [female_count, male_count]
    labels = ['Female', 'Male']
    
    # Generate pastel color palette.
    pastel_colors = generate_pastel_palette(6)
    
    # Define labels and sizes based on the counts.
    labels = []
    sizes = []
    if male_count > 0:
        labels.append('Male')
        sizes.append(male_count)
    if female_count > 0:
        labels.append('Female')
        sizes.append(female_count)    

    # Create legend labels with additional text.
    legend_labels = [f'{label} ({count})' for label, count in zip(labels, sizes)]
    
    # Set explode only if labels includes both genders.
    if len(labels) > 1:
        explode = (0.018, 0.018)
    else:
        explode = None

    # Create the pie chart.
    wedges, texts, autotexts = ax.pie(sizes, 
                                      #colors=[pastel_colors[0], pastel_colors[1]], 
                                      colors=[pastel_colors[2] if label == 'Female' else pastel_colors[5] for label in labels],
                                      labels=labels, 
                                      autopct='%1.1f%%', 
                                      pctdistance=0.79, 
                                      explode=explode,
                                      labeldistance=1.125, 
                                      wedgeprops={'linewidth': 0.3, 'edgecolor': pie_wedge_color},
                                      textprops={'color': annotation_color, 'weight': 'normal', 'fontsize': '10'},
                                      shadow=True)

    # Draw inner circle for donut effect.
    centre_circle = plt.Circle((0, 0), 0.65, fc=chart_bg_color)
    ax.add_artist(centre_circle)

    # Set title of the chart.
    #ax.set_title(f' ${salary_min:,} - ${salary_max:,}', fontsize=10)  # Formatting with dollar signs and comma separators
    ax.set_title('{:,.0f} - {:,.0f}'.format(salary_min, salary_max), fontsize=12, color='#966FD6', fontweight='bold')
    
    # Set aspect ratio to be equal.
    ax.set_aspect('equal')
    
    # Set font color of text labels
    for text in texts:
        text.set_color('black')
        text.set_fontsize('9')

# Create subplots for the two pie charts
#fig, (ax1, ax2, ax3, ax4, ax5) = plt.subplots(1, 5, figsize=(20, 14))
fig, ((ax1, ax2, ax3), (ax4, ax5, ax6)) = plt.subplots(2, 3, figsize=(16, 12), gridspec_kw={'hspace': 0.1})

# Call the function to create the first pie chart for the salary range $45k - $75k
create_salary_range_pie(ax1, df, 45000, 74999)

# Call the function to create the first pie chart for the salary range $75k - $100k
create_salary_range_pie(ax2, df, 75000, 99999)

# Call the function to create the second pie chart for the salary range $100k - $130k
create_salary_range_pie(ax3, df, 100000, 129999)

# Call the function to create the second pie chart for the salary range $130k - $180k
create_salary_range_pie(ax4, df, 130000, 179999)

# Call the function to create the second pie chart for the salary range $180k+
create_salary_range_pie(ax5, df, 180000, df['Salary'].max())

# Hide the 6th subplot
ax6.axis('off')

# Add main title for all subplots.
fig.suptitle('Distribution of Gender by Specific Salary Range', fontsize=12, fontweight='bold')

# Set background color for the entire chart.
fig.set_facecolor(chart_bg_color)

# Adjust the position of the subplots in the second row by adding padding.
for ax in [ax4, ax5, ax6]:
    pos = ax.get_position()
    pos.x0 += 0.14
    pos.x1 += 0.14
    ax.set_position(pos)

# Adjust layout and display the charts with reduced space for the super title.
plt.tight_layout(rect=[0, 0, 1, 0.05])
plt.show()

# Group the dataframe by both 'Position' and 'Sex', calculate the mean salary for each group.
mean_salary_by_position_sex = df_positions.groupby(['Position', 'Sex'])['Salary'].mean()#.sort_values(ascending=False)

# Convert the sorted Series to a dataframe and reset the index
mean_salary_by_position_sex_df = mean_salary_by_position_sex.reset_index()

# Group the dataframe by 'Position' and count the unique values in 'Sex'
position_sex_counts = df.groupby('Position')['Sex'].nunique()

# Select only positions where both male and female are present.
positions_with_both_sexes = position_sex_counts[position_sex_counts == 2].index

# Filter the original dataframe based on positions with both sexes.
mean_salary_by_position_sex_df_both_sexes = mean_salary_by_position_sex_df[mean_salary_by_position_sex_df['Position'].isin(positions_with_both_sexes)]

# Remove leading and trailing whitespaces from 'Sex' column.
mean_salary_by_position_sex_clean = mean_salary_by_position_sex_df_both_sexes.copy()
mean_salary_by_position_sex_clean['Sex'] = mean_salary_by_position_sex_df_both_sexes['Sex'].str.strip()
mean_salary_by_position_sex_clean

# Create a pivot table with mean salary for each position grouped by gender
pivot_table = mean_salary_by_position_sex_clean.pivot_table(index='Position', columns='Sex', values='Salary', aggfunc='mean')

# Sort the pivot table based on the mean salary for a specific gender group (e.g., male)
sorted_pivot_table = pivot_table.sort_values(by='M', ascending=True)

# Generate a pastel color palette.
pastel_colors = generate_pastel_palette(6)
colors = [pastel_colors[2] if label == 'F' else pastel_colors[5] for label in sorted_pivot_table.columns]

# Create legend labels with 'Female' instead of 'F' and 'Male' instead of 'M'
legend_labels = [label.replace('F', 'Female').replace('M', 'Male') for label in sorted_pivot_table.columns]

# Plot a stacked bar chart.
ax = sorted_pivot_table.plot(kind='bar', 
                             stacked=True, # Stacked bar chart.
                             figsize=(15, 6), 
                             color=colors, 
                             edgecolor=bar_edge_color, 
                             linewidth=0.5)

# Hide spines.
ax.spines['top'].set_visible(False) 
ax.spines['right'].set_visible(False)

# Chart background color.
ax.set_facecolor(chart_bg_color)

# Get the current tick positions and labels
ticks = ax.get_xticklabels()

# Format the tick values y as a string with dollar sign, comma separator, and zero decimal places.
yticks = np.arange(0, max_ytick + 1, 25000)
ytick_labels = ['${:,.0f}'.format(y) for y in yticks]

# Set yticks and yticklabels
ax.set_yticks(yticks)
ax.set_yticklabels(ytick_labels)

# Adjust the position of the tick labels slightly to the left
for tick in ticks:
    tick.set_horizontalalignment('right')
    tick.set_x(-0.5)
    
# Calculate the percentage for each row in the pivot table
total = pivot_table.sum(axis=1)  # Get the total for each row
per = pivot_table.div(total, axis=0).mul(100).round(2)  # Calculate the percent for each row
    
# Custom annotation labels for stacked bars
for c in ax.containers:
    
    # Get the current segment label (a string); corresponds to column / legend
    label = c.get_label()
    
    # Create custom labels with the percentage from the per DataFrame.
    # The column labels are strings.
    if label == 'F':
        column = 'F'
    else:
        column = 'M'
    
    # Add the annotation for each segment in the container
    for i, bar in enumerate(c):
        height = bar.get_height()
        if height != 0:  # Skip annotating if the bar height is 0
            ax.annotate(f'{per.loc[sorted_pivot_table.index[i], column]}%',  # Annotation text
                        xy=(bar.get_x() + bar.get_width() / 2, bar.get_y() + bar.get_height()),  # Annotation position
                        xytext=(0, 3),  # Offset for the annotation text
                        textcoords="offset points",
                        ha='center', va='bottom',  # Text alignment
                        fontsize=8, fontweight='normal', color=annotation_color)  # Font properties
            
# Set the title.
plt.title('Total Mean Salary by Position and Gender', fontweight='bold', fontsize=14)
# Add the subtitle.
plt.text(0.5, 0.98, 'Includes only Positions with both genders', horizontalalignment='center', verticalalignment='center',fontsize=10, transform=ax.transAxes)

plt.xlabel('')
plt.ylabel('Total Mean Salary', fontweight='bold')

# Grid lines
plt.grid(axis='y', linestyle='--', alpha=0.15)

# Show legend.
legend = plt.legend(legend_labels, loc="upper left", bbox_to_anchor=(0.05, 1.01))
legend.set_title('Gender')
legend.get_title().set_fontweight('bold')

# Show the plot
plt.xticks(rotation=45)  # Rotate x-axis labels for better readability
plt.tight_layout()
plt.show()

# Select Network Engineers from dataframe.
network_engineer_df = df_positions[df_positions['Position'] == 'Network Engineer']

# Count the number of females and males.
female_count = (network_engineer_df['Sex'] == 'F').sum()
male_count = (network_engineer_df['Sex'] == 'M').sum()

# Display the counts.
print("Count of Females as Network Engineers:", female_count)
print("Count of Males as Network Engineers:", male_count)

Count of Females as Network Engineers: 2
Count of Males as Network Engineers: 3

# Generate a pastel color palette.
pastel_colors = generate_pastel_palette(6)
colors = [pastel_colors[2] if label == 'F' else pastel_colors[5] for label in sorted_pivot_table.columns]

# Plot grouped bar chart
ax = pivot_table[['F', 'M']].plot(kind='bar', 
                                  figsize=(12, 6), 
                                  width=0.35, 
                                  color=colors,
                                  edgecolor=bar_edge_color, 
                                  linewidth=0.5)

# Define the category you want to highlight
category_to_highlight = "Network Engineer"

# Get the index of the category in the dataframe.
index_to_highlight = pivot_table.index.get_loc(category_to_highlight)

# Set chart background color
ax.set_facecolor(chart_bg_color)

# Format the tick values y as a string with dollar sign, comma separator, and zero decimal places.
yticks = np.arange(0, max_ytick + 1, 10000)
ytick_labels = ['${:,.0f}'.format(y) for y in yticks]

# Set yticks and yticklabels
ax.set_yticks(yticks)
ax.set_yticklabels(ytick_labels)

# Hide spines.
ax.spines['top'].set_visible(False) 
ax.spines['right'].set_visible(False)

plt.rcParams['axes.facecolor'] = 'lightgrey'
# Get the current tick positions and labels
ticks = ax.get_xticklabels()
# Adjust the position of the tick labels slightly to the left
for tick in ticks:
    tick.set_horizontalalignment('right')
    tick.set_x(-0.5)  # Adjust this value as needed for the desired amount of movement

# Set title and labels
plt.title('Mean Salary by Position and Gender', fontweight='bold')
plt.xlabel('')
plt.ylabel('Mean Salary', fontweight='bold')
plt.xticks(rotation=45)

# Add the subtitle.
plt.text(0.5, 0.98, 'Includes only Positions with both genders', horizontalalignment='center', verticalalignment='center', fontsize=10, transform=ax.transAxes)

# Add grid lines with light grey color
plt.grid(axis='y', linestyle='--', color='lightgrey', alpha=0.35)

# Show legend
legend = plt.legend(labels=['Female', 'Male'], loc="upper right", bbox_to_anchor=(1.15, 1.01))
legend.set_title('Gender')
legend.get_title().set_fontweight('bold')

# Add annotation to the top of the "network engineer" bar
bar_rects = ax.patches
for i, rect in enumerate(bar_rects):
    if i == index_to_highlight:
        x = rect.get_x() + rect.get_width() / 2
        y = rect.get_height()
        ax.annotate('Salary Disparity:\nFemale Network Engineer', xy=(x, y), xytext=(0, 70), textcoords='offset points',
                    ha='center', va='bottom', fontsize=10, color='darkblue',
                    arrowprops=dict(arrowstyle='->', connectionstyle='arc3', color='darkblue'))

# Show plot
plt.tight_layout()
plt.ylim(0,130000)
plt.grid(True, alpha=0.15)
plt.show()

# Mean attrition rate by Age.
attritionrate_by_age = df.groupby('Age')['Termd'].mean().reset_index(name='AttritionRate')
attritionrate_by_age.sort_values(by=['AttritionRate', 'Age']).tail(7)

pearson_coef, p_value = stats.pearsonr(df['Age'], df['Termd'])
print("The Pearson Correlation Coefficient is", pearson_coef, " with a P-value of P =", p_value)

The Pearson Correlation Coefficient is 0.1019948649696528  with a P-value of P = 0.07247178632527944

# Generate a pastel color palette.
pastel_colors = generate_pastel_palette(2)

# Marker data point size
marker_size = 45

# Create a figure window and axes to plot data.
fig, ax = plt.subplots(figsize=(8,6))

# Set chart background colour.
ax.set_facecolor(chart_bg_color)

# Plot the regression plot for Mean Attrition Rate by Age.
sns.regplot(x='Age', y='AttritionRate', data=attritionrate_by_age, 
            scatter=True,
            color=pastel_colors[0],
            scatter_kws={'s': marker_size, 'edgecolor': scatter_edge_color},
            line_kws={"color":"red"},
            ci=95) # confidence interval

# Set the title and labels
plt.title('Average Attrition Rate by Age', **title_label_params)
plt.xlabel('Age', fontweight='bold', labelpad=10)
plt.ylabel('', fontweight='bold', labelpad=10)

# Manually adjust y-axis tick labels to percentages.
y_ticks = ax.get_yticks()
ax.set_yticklabels(['{:,.0%}'.format(y) for y in y_ticks])

plt.xticks(range(30, 71, 5))

# Show the plot
plt.grid(True, linestyle='--', alpha=0.76)
plt.tight_layout()
plt.show()

# Determine the starting point for the age bins.
min_age = df['Age'].min()

# Round down the minimum age to the nearest multiple of 5
min_age_rounded = min_age - (min_age % 5)

# Define the bins for age.
bins = np.arange(min_age_rounded, df['Age'].max() + 5, 5)

grouped_by_age_bins = pd.cut(df['Age'], bins)

# Create age bins and calculate mean attrition rate for each bin
grouped_by_age_bins = df.groupby(grouped_by_age_bins)['Termd'].mean().reset_index(name='AttritionRate')

# Display the dataframe.
print(grouped_by_age_bins)

        Age  AttritionRate
0  (30, 35]       0.280000
1  (35, 40]       0.321839
2  (40, 45]       0.295775
3  (45, 50]       0.326531
4  (50, 55]       0.432432
5  (55, 60]       0.320000
6  (60, 65]       0.200000
7  (65, 70]       0.500000
8  (70, 75]       0.666667

# Generate a pastel color palette.
pastel_colors = generate_pastel_palette(len(grouped_by_age_bins))

# Extract left edge of each interval as bin labels
bin_labels = [str(interval.left) for interval in grouped_by_age_bins['Age']]

# Create the bar graph
fig, ax = plt.subplots(figsize=(8, 6))
plt.bar(bin_labels, grouped_by_age_bins['AttritionRate'], 
        color=pastel_colors,
        edgecolor=bar_edge_color,
        linewidth=0.5)

# Define x-values for the trendline
x = np.arange(len(bin_labels))

# Add trendline for attrition rate
z = np.polyfit(x, grouped_by_age_bins['AttritionRate'], 1)
p = np.poly1d(z)
ax.plot(x, p(x), color='red', linestyle='--', label='Attrition Rate Trendline')

# Annotate the trendline with the slope
slope = z[0]
text_x = len(bin_labels) * 0.5 + 0.7 # Positioning the text in the middle of the x-axis
text_y = p(text_x + 1)  # Positioning the text on the trendline
ax.text(text_x, text_y, f'Slope: {slope:.2f}', color='red', fontsize=12, ha='center', va='center')

# Manually adjust y-axis tick labels to percentages.
y_ticks = ax.get_yticks()
ax.set_yticklabels(['{:,.0%}'.format(y) for y in y_ticks])

# Set chart background colour.
ax.set_facecolor(chart_bg_color)

# Hide spines.
ax.spines['top'].set_visible(False) 
ax.spines['right'].set_visible(False)

# Set the title and labels
plt.title('Average Attrition Rate by Age Bins', **title_label_params)
plt.xlabel('Age Bins', fontweight='bold', labelpad=10)
plt.ylabel('Attrition Rate', fontweight='bold', labelpad=10)

# Show the plot
plt.grid(True, linestyle='--', alpha=0.65, axis='y')
plt.xticks(rotation=0)
plt.tight_layout()
plt.show()

# Create age bins and calculate mean attrition rate for each bin.
grouped_by_department = df.groupby('Department')['Termd'].mean().reset_index(name='AttritionRate')

# Convert attrition rate to percentage by multiplying by 100
grouped_by_department['AttritionRate'] *= 100
grouped_by_department.round(2)

# Filter the dataframe for terminated employees only.
df_employed = df[df['TermReason'] != 'N/A-StillEmployed']

# Create a pivot table to count the occurrences of each race for each recruitment source
dept_counts_per_reason_df = df_employed.pivot_table(index='Department', columns='TermReason', aggfunc='size', fill_value=0)

# Reset the index and rename the columns
dept_counts_per_reason_df = dept_counts_per_reason_df.reset_index().rename_axis(columns=None).set_index('Department')
dept_counts_per_reason_df

# Generate a pastel color palette.
pastel_colors = generate_pastel_palette(len(dept_counts_per_reason_df.columns))

# Exclude the 'Department' column from the sum calculation
dept_counts_per_reason_df['Total'] = dept_counts_per_reason_df.sum(axis=1)

# Calculate the proportion of each termination reason for each department
proportion_dept_counts_df = dept_counts_per_reason_df.div(dept_counts_per_reason_df['Total'], axis=0)
proportion_dept_counts_df.drop('Total', axis=1, inplace=True)

# Custom bar width.
bar_width = 0.75

# Plot the bar plot
fig, ax = plt.subplots(figsize=(8, 6))
proportion_dept_counts_df.plot(
    kind='bar', 
    color=pastel_colors, 
    edgecolor=bar_edge_color, 
    linewidth=0.5, 
    width=bar_width, 
    stacked=True, 
    ax=ax
)
ax.set_facecolor(chart_bg_color)

# Plotting
plt.title('Proportion of Termination Reasons by Department', **title_label_params)
plt.xlabel('')
plt.ylabel('Proportion', fontweight='bold', labelpad=10)
plt.xticks(rotation=50, ha='right')
plt.legend(title='Termination Reason', bbox_to_anchor=(1.04, 1), loc='upper left')
plt.tight_layout()
plt.show()

# Define the mapping of individual termination reasons to broader categories.
reason_to_category = {
    'Another position': 'Personal Reasons',
    'Fatal attraction': 'Personal Reasons',
    'Learned that he is a gangster': 'Personal Reasons',
    'attendance': 'Attendance',
    'career change': 'Career Development',
    'gross misconduct': 'Performance',
    'hours': 'Performance',
    'maternity leave - did not return': 'Health-related',
    'medical issues': 'Health-related',
    'military': 'Career Development',
    'more money': 'Financial Considerations',
    'no-call, no-show': 'Attendance',
    'performance': 'Performance',
    'relocation out of area': 'Relocation',
    'retiring': 'Retiring',
    'return to school': 'Career Development',
    'unhappy': 'Personal Reasons'
}

# Apply the mapping to create a new column with broader categories.
df_employed['TermReason_Category'] = df_employed['TermReason'].map(reason_to_category)

# Create a pivot table to count the occurrences of each race for each recruitment source
dept_counts_per_reason_df = df_employed.pivot_table(index='Department', columns='TermReason_Category', aggfunc='size', fill_value=0)

# Reset the index and rename the columns
dept_counts_per_reason_df = dept_counts_per_reason_df.reset_index().rename_axis(columns=None).set_index('Department')
dept_counts_per_reason_df

# Generate a pastel color palette.
pastel_colors = generate_pastel_palette(len(dept_counts_per_reason_df.columns))

# Total reasons per department.
dept_counts_per_reason_df['Total'] = dept_counts_per_reason_df.sum(axis=1)

# Calculate the proportion of each termination reason for each department.
proportion_dept_counts_df = dept_counts_per_reason_df.div(dept_counts_per_reason_df['Total'], axis=0)

# Drop the 'Total' column.
proportion_dept_counts_df.drop('Total', axis=1, inplace=True)

# Custom bar width.
bar_width = 0.75

# Plot the bar plot.
fig, ax = plt.subplots(figsize=(8, 6))
proportion_dept_counts_df.plot(
    kind='bar', 
    color=pastel_colors, 
    edgecolor=bar_edge_color, 
    linewidth=0.5, 
    width=bar_width, 
    stacked=True, 
    ax=ax
)
ax.set_facecolor(chart_bg_color)

# Plotting
plt.title('Proportion of Termination Reasons by Department', **title_label_params)
plt.xlabel('')
plt.ylabel('Proportion', fontweight='bold', labelpad=10)
plt.xticks(rotation=50, ha='right')
plt.legend(title='Termination Reason', bbox_to_anchor=(1.04, 1), loc='upper left')
plt.tight_layout()
plt.show()

# Copy original dataframe.
race_df = df.copy()

# Group by RaceDesc and calculate the mean performance score and attrition rate.
race_stats = race_df.groupby('RaceDesc').agg({'Termd': 'mean'}).reset_index()

# Sort by performance score in descending order.
race_stats_sorted = race_stats.sort_values(by='Termd', ascending=True)
race_stats_sorted

# Generate a pastel color palette.
pastel_colors = generate_pastel_palette(6)

# Plotting
fig, ax = plt.subplots(figsize=(10, 6), facecolor='lightblue')

# Set background color of plot area.
ax.set_facecolor(chart_bg_color)

# Plot the data
ax.plot(race_stats_sorted['RaceDesc'], race_stats_sorted['Termd'], marker='o', 
        color=pastel_colors[2], linestyle='-',
        markersize=8, 
        linewidth=3.0)

# Add labels and title
ax.set_title('Attrition Rate by Race Description', color='black', **title_label_params)
ax.set_xlabel('Race Description', color='black', fontweight='bold')
ax.set_ylabel('Attrition Rate', color='black', fontweight='bold', labelpad=15)

# Set y-tick labels colour.
plt.yticks(color='black')

# Set y-axis limits.
plt.ylim(-0.02, 0.40)

# Convert y-axis values to percentages
y_vals = ax.get_yticks()
ax.set_yticklabels([f'{int(y * 100)}%' for y in y_vals])

# Adjust x-tick labels to add a new line for each word.
ax.set_xticklabels([label.replace(" ", "\n") for label in race_stats_sorted['RaceDesc']], color='black', ha='center', fontsize=9)

# Annotate markers with percentages rounded to two decimal places
for i, (x, y) in enumerate(zip(race_stats_sorted['RaceDesc'], race_stats_sorted['Termd'])):
    ax.annotate(f'{y * 100:.2f}%', (x, y), textcoords="offset points", xytext=(-2,9), ha='center', color='darkblue')

# Make grid lines thinner.
ax.grid(True, linewidth=1.0, linestyle=':', alpha=0.80)

plt.grid(True)
plt.tight_layout()
plt.show()

# Calculate the total count of employees.
total_employee_count = race_df.shape[0]

# Group by 'RaceDesc' and calculate the sum of 'Termd' column.
overall_race_stats = race_df.groupby('RaceDesc').agg({'Termd': 'sum'}).reset_index()

# Calculate the attrition rate relative to the total count of people in the organisation for each race.
overall_race_stats['AttritionRate'] = overall_race_stats['Termd'] / total_employee_count
overall_race_stats

# Sort the dataframe by attrition rate in ascending order.
race_stats_sorted = overall_race_stats.sort_values(by='Termd')

# Generate a pastel color palette.
pastel_colors = generate_pastel_palette(6)

# Plotting
fig, ax = plt.subplots(figsize=(10, 6), facecolor='lightblue')

# Set background color of plot area.
ax.set_facecolor(chart_bg_color)

# Plot the data
ax.plot(race_stats_sorted['RaceDesc'], race_stats_sorted['AttritionRate'], marker='o', 
        color=pastel_colors[2], linestyle='-',
        markersize=8, 
        linewidth=3.0)

# Add labels and title
ax.set_title('Race Attrition Impact on Total Workforce', color='black', **title_label_params)
ax.set_xlabel('Race Description', color='black', fontweight='bold')
ax.set_ylabel('Attrition Rate', color='black', fontweight='bold', labelpad=15)

# Set y-tick labels colour.
plt.yticks(color='black')

# Set y-axis limits.
plt.ylim(-0.01, 0.23)

# Convert y-axis values to percentages
y_vals = ax.get_yticks()
ax.set_yticklabels([f'{int(y * 100)}%' for y in y_vals])

# Adjust x-tick labels to add a new line for each word.
ax.set_xticklabels([label.replace(" ", "\n") for label in race_stats_sorted['RaceDesc']], color='black', ha='center', fontsize=9)

# Annotate markers with percentages rounded to two decimal places
for i, (x, y) in enumerate(zip(race_stats_sorted['RaceDesc'], race_stats_sorted['AttritionRate'])):
    ax.annotate(f'{y * 100:.2f}%', (x, y), textcoords="offset points", xytext=(-2,9), ha='center', color='darkblue')

# Make grid lines thinner.
ax.grid(True, linewidth=1.0, linestyle=':', alpha=0.80)

plt.grid(True)
plt.tight_layout()
plt.show()

race_term_counts = df_employed.groupby(['RaceDesc','TermReason']).size().reset_index(name='Count')

# Get top 3 termination reasons for each race
top_3_reasons_per_race = race_term_counts.groupby('RaceDesc').apply(lambda x: x.nlargest(3, 'Count')).reset_index(drop=True)
top_3_reasons_per_race

pearson_coef, p_value = stats.pearsonr(df['SpecialProjectsCount'], df['Termd'])
print("The Pearson Correlation Coefficient is", pearson_coef, " with a P-value of P =", p_value)

The Pearson Correlation Coefficient is -0.14742910417277896  with a P-value of P = 0.009221967246014609

# Calculate the mean attrition rate for each SpecialProjectsCount group
attrition_rate_by_project_count = df.groupby('SpecialProjectsCount')['Termd'].mean()

# Generate a pastel color palette.
pastel_colors = generate_pastel_palette(len(attrition_rate_by_project_count))

bar_width = 0.7 # custom bar width.

# Plot the bar plot.
fig, ax = plt.subplots(figsize=(8,6))
ax.set_facecolor(chart_bg_color)

# Hide spines
ax.spines['top'].set_visible(False)
ax.spines['right'].set_visible(False)

# Plotting the attrition rate for each SpecialProjectsCount
attrition_rate_by_project_count.plot(kind='bar', color=pastel_colors, edgecolor=bar_edge_color, linewidth=0.5, width=bar_width)

# Annotate bars with percentages
for i, rate in enumerate(attrition_rate_by_project_count):
    ax.text(i, rate + 0.01, f'{rate:.1%}', ha='center', color='darkblue', fontsize=10)

# Set y-label ticks to display percentages
ax.set_yticklabels([f'{tick:.0%}' for tick in ax.get_yticks()])

# Customising the plot
plt.xlabel('Special Projects Count', fontweight='bold', labelpad=10)
plt.ylabel('Attrition Rate', fontweight='bold', labelpad=10)
plt.ylim(0.0, 0.6)
plt.title('Mean Attrition Rate by Special Projects Count', **title_label_params)
plt.xticks(rotation=0)  # Ensure the x-axis labels are not rotated
plt.grid(True, alpha=0.60, axis='y', linestyle=':')
plt.show()

# Convert 'DateofTermination' column to datetime.
df['DateofTermination'] = pd.to_datetime(df['DateofTermination'])

# Extract year from TerminationYear.
df['TerminationYear'] = df['DateofTermination'].dt.year

# Count occurrences of each year.
termination_counts = df['TerminationYear'].value_counts()

# Display year and total count for terminations.
print("Year   TotalCount")
print("-----------------")
for year, count in termination_counts.sort_index().items():
    print(f"{int(year)}    {count}")

Year   TotalCount
-----------------
2010    1
2011    3
2012    8
2013    13
2014    13
2015    23
2016    22
2017    8
2018    13

# Filter to terminations in 2015 and 2016.
df_terminations_2015_2016 = df[(df['DateofTermination'].dt.year.isin([2015, 2016])) & (df['Termd'] == 1)]

# Group by TermReason and count terminations for each reason
reason_termination_counts = df_terminations_2015_2016['TermReason'].value_counts()

# Display the reasons for terminations in 2015 and 2016.
print("Reasons for terminations in 2015 and 2016:\n")
print(reason_termination_counts.head(7))

Reasons for terminations in 2015 and 2016:

TermReason
unhappy             6
Another position    5
more money          4
attendance          4
career change       3
performance         3
hours               3
Name: count, dtype: int64

# Group by ManagerName and count terminations for each manager.
manager_termination_counts = df_terminations_2015_2016[['ManagerName','Department']].value_counts()

# Display the manager with the most terminations in 2015 and 2016.
print("Manager with the most terminations in 2015 and 2016:\n")
manager_termination_counts.head()

Manager with the most terminations in 2015 and 2016:

ManagerName     Department
Webster Butler  Production    7
Amy Dunn        Production    7
Simon Roup      IT/IS         6
Elijiah Gray    Production    4
Kissy Sullivan  Production    3
Name: count, dtype: int64

# Convert 'DateofHire' and 'DateofTermination' columns to datetime format.
yos_df = df.copy()
yos_df['DateofHire'] = pd.to_datetime(yos_df['DateofHire'])
yos_df['DateofTermination'] = pd.to_datetime(yos_df['DateofTermination'])

# Get the current date.
current_date = pd.to_datetime('now')

# Calculate the duration of service for active employees.
yos_df['DateofTermination'] = yos_df['DateofTermination'].fillna(current_date, inplace=False)
yos_df['years_of_service'] = ((yos_df['DateofTermination'] - yos_df['DateofHire']).dt.days / 365.25).astype(int) # account for leap years

# Define bins for years of service
bins = np.arange(0, yos_df['years_of_service'].max(), 5)

# Create labels for the bins. Bin edges are inclusive on the left and exclusive on the right. 
labels = [f"{i}-{i+4}" for i in bins[:-1]]

# Convert years_of_service into bins.
yos_df['years_of_service_bins'] = pd.cut(yos_df['years_of_service'], bins=bins, labels=labels, right=False)

# Group by years of service bins and termination status, then calculate the mean
attrition_rate_yos = yos_df.groupby('years_of_service_bins')['DateofTermination'].apply(lambda x: (x < current_date).mean()).reset_index()
attrition_rate_yos.rename(columns={'DateofTermination': 'attrition_rate'}, inplace=True)

print(attrition_rate_yos)

  years_of_service_bins  attrition_rate
0                   0-4        1.000000
1                   5-9        0.220339
2                 10-14        0.000000

# Generate a pastel color palette.
pastel_colors = generate_pastel_palette(yos_df['years_of_service_bins'].nunique())

bar_width = 0.7 # custom bar width.

# Plot the bar plot.
fig, ax = plt.subplots(figsize=(6,5))
ax.set_facecolor(chart_bg_color)

# Hide spines
ax.spines['top'].set_visible(False)
ax.spines['right'].set_visible(False)

plt.bar(attrition_rate_yos['years_of_service_bins'], attrition_rate_yos['attrition_rate'], 
        color=pastel_colors, edgecolor=bar_edge_color, linewidth=0.5, width=bar_width)

# Set y-label ticks to display percentages
yticks = np.arange(0, 1.01, 0.25)
ax.set_yticks(yticks)
ax.set_yticklabels([f'{tick:.0%}' for tick in yticks])

# Adding labels and title
plt.title('Attrition Rate by Years of Service', **title_label_params)
plt.xlabel('Years of Service', fontweight='bold', labelpad=10)
plt.ylabel('Attrition Rate', fontweight='bold', labelpad=10)
plt.xticks(rotation=0)

# Display plot
plt.tight_layout()
plt.grid(True, alpha=0.60, axis='y', linestyle=':')
plt.show()

# Group by 'Salary' and calculate the mean of 'Termd'
attrition_by_salary = df.groupby('Salary')['Termd'].mean().reset_index(name='AttritionRate')

# Define the bins for Salary.
bin_step = 20000
bins = np.arange(40000, attrition_by_salary['Salary'].max(), bin_step)

# Create labels for the bins. Bin edges are inclusive on the left and exclusive on the right. 
labels = [f"{i}-{i+bin_step}" for i in bins[:-1]]

# Convert Salary into bins.
grouped_by_salary_bins = pd.cut(df['Salary'], bins=bins, labels=labels, right=False)

# Create salary bins and calculate mean attrition rate for each bin
attrition_by_salary_bins = df.groupby(grouped_by_salary_bins)['Termd'].mean().reset_index(name='AttritionRate')

# Fill missing values in AttritionRate column with 0.
attrition_by_salary_bins['AttritionRate'].fillna(0, inplace=True)

# Display the DataFrame
print(attrition_by_salary_bins)

          Salary  AttritionRate
0    40000-60000       0.404959
1    60000-80000       0.295455
2   80000-100000       0.212121
3  100000-120000       0.500000
4  120000-140000       0.500000
5  140000-160000       0.250000
6  160000-180000       0.000000
7  180000-200000       0.000000
8  200000-220000       0.000000
9  220000-240000       0.000000

# Generate a pastel color palette.
pastel_colors = generate_pastel_palette(attrition_by_salary_bins['Salary'].nunique())

bar_width = 0.8 # custom bar width.

# Plot the bar plot.
fig, ax = plt.subplots(figsize=(6,5))
ax.set_facecolor(chart_bg_color)

# Hide spines
ax.spines['top'].set_visible(False)
ax.spines['right'].set_visible(False)

plt.bar(attrition_by_salary_bins['Salary'], attrition_by_salary_bins['AttritionRate'], 
        color=pastel_colors, edgecolor=bar_edge_color, linewidth=0.5, width=bar_width)

# Set y-label ticks to display percentages
yticks = np.arange(0, 1.01, 0.10)
ax.set_yticks(yticks)
ax.set_yticklabels([f'{tick:.0%}' for tick in yticks])

# Adding labels and title
plt.title('Attrition Rate by Salary Bins', **title_label_params)
plt.xlabel('Salary Bins', fontweight='bold', labelpad=10)
plt.ylabel('Attrition Rate', fontweight='bold', labelpad=10)
plt.xticks(rotation=45)

# Get the current tick positions and labels.
ticks = ax.get_xticklabels()

# Adjust the position of the tick labels slightly to the left.
for tick in ticks:
    tick.set_horizontalalignment('right')
    tick.set_x(-0.5)
    
# Modify the x-tick labels to add $ and comma separators.
new_tick_labels = []
for tick in ticks:
    salary_range = tick.get_text()
    # Add $ and format with commas
    new_label = '${:,.0f}'.format(float(salary_range.split('-')[0])) + '-' + '{:,.0f}'.format(float(salary_range.split('-')[1]))
    new_tick_labels.append(new_label)

# Set the modified tick labels
plt.xticks(np.arange(len(new_tick_labels)), new_tick_labels, rotation=45)

# Display plot
plt.tight_layout()
plt.ylim(0, 0.70)
plt.grid(True, alpha=0.70, axis='y', linestyle=':')
plt.show()

# Group by 'ManagerName' and count the number of unique departments.
manager_departments_count = df.groupby('ManagerName')['Department'].nunique()

# Filter managers who have worked in more than one department.
managers_multiple_depts = manager_departments_count[manager_departments_count > 1].index.tolist()

# Filter the original dataframe to include only the rows where the manager has worked in multiple departments.
df_multiple_depts = df[df['ManagerName'].isin(managers_multiple_depts)]

print(f'Managers who have worked in multiple departments: {managers_multiple_depts}')

Managers who have worked in multiple departments: ['Board of Directors', 'Janet King', 'Jennifer Zamora']

# Group by Manager and calculate the mean attrition rate.
manager_attrition_rates = df.groupby(['ManagerName'])['Termd'].mean().reset_index().sort_values(by='Termd', ascending=False)
manager_attrition_rates = manager_attrition_rates.rename(columns={'Termd': 'AttritionRate'})
manager_attrition_rates.head()

# Generate a pastel color palette.
pastel_colors = generate_pastel_palette(len(manager_attrition_rates))

bar_width = 0.9 # custom bar width.

# Plot the bar plot.
fig, ax = plt.subplots(figsize=(10,6))
ax.set_facecolor(chart_bg_color)

# Set y-label ticks to display percentages
yticks = np.arange(0, 1.01, 0.10)
ax.set_yticks(yticks)
ax.set_yticklabels([f'{tick:.0%}' for tick in yticks])

# Get the current tick positions and labels.
ticks = ax.get_xticklabels()

# Adjust the position of the tick labels slightly to the left.
for tick in ticks:
    tick.set_horizontalalignment('right')
    tick.set_x(-0.5)

# Hide spines
ax.spines['top'].set_visible(False)
ax.spines['right'].set_visible(False)

# Plot the bar plot
plt.bar(manager_attrition_rates['ManagerName'], manager_attrition_rates['AttritionRate'], 
        color=pastel_colors, edgecolor=bar_edge_color, linewidth=0.5, width=bar_width)

# Annotate bars with percentages
for i, rate in enumerate(manager_attrition_rates['AttritionRate']):
    if rate > 0.00:
        ax.text(i, rate + 0.01, f'{rate:.0%}', ha='center', color='darkblue', fontsize=9)

# Add labels and title
plt.xlabel('', fontsize=12)
plt.ylabel('Attrition Rate', fontweight='bold', labelpad=10)
plt.title('Manager-wise Attrition Rates', **title_label_params)
plt.xticks(rotation=45)

# Display plot
plt.tight_layout()
plt.ylim(0, 0.70)
plt.grid(True, alpha=0.70, axis='y', linestyle=':')
plt.show()

# Filter rows where ManagerName is 'Webster Butler'
webster_butler_data = df[df['ManagerName'] == 'Webster Butler']

# Calculate the sum of the Terminations.
total_termds = webster_butler_data['Termd'].sum()

# Count the number of rows for the total number of employees.
total_employees = len(webster_butler_data)

print("Total Terminations for Webster Butler:", total_termds)
print("Total Employees under Webster Butler:", total_employees)

Total Terminations for Webster Butler: 13
Total Employees under Webster Butler: 21

# Group the dataframe by ManagerName and calculate the mean engagement survey score and attrition rate.
engagement_attrition_manager = df.groupby('ManagerName').agg({'Termd': 'mean', 'EngagementSurvey': 'mean'}).reset_index().rename(columns={'Termd': 'AttritionRate'})

# Sort by performance score in descending order
engagement_attrition_manager = engagement_attrition_manager.sort_values(by='AttritionRate', ascending=False)
engagement_attrition_manager.head()

# Check Janet King's department in engagement_attrition_manager
janet_king_in_performance = engagement_attrition_manager[engagement_attrition_manager['ManagerName'] == 'Janet King']
janet_king_in_performance

# Calculate Pearson correlation coefficient and p-value for EmpSatisfaction vs Termd
pearson_coef_emp, p_value_emp = stats.pearsonr(df['EmpSatisfaction'], df['Termd'])
print("The Pearson Correlation Coefficient between EmpSatisfaction and Termd is", pearson_coef_emp, " with a P-value of P =", p_value_emp)

# Calculate Pearson correlation coefficient and p-value for EngagementSurvey vs Termd
pearson_coef_eng, p_value_eng = stats.pearsonr(df['EngagementSurvey'], df['Termd'])
print("The Pearson Correlation Coefficient between EngagementSurvey and Termd is", pearson_coef_eng, " with a P-value of P =", p_value_eng)

The Pearson Correlation Coefficient between EmpSatisfaction and Termd is -0.004731666973705058  with a P-value of P = 0.9337654152178527
The Pearson Correlation Coefficient between EngagementSurvey and Termd is -0.01754287622809008  with a P-value of P = 0.7579682377443097

terminated_female_count = ((df['Sex'] == 'F') & (df['Termd'] == 1)).sum()
print(f'Total females terminated: {terminated_female_count}')

terminated_male_count = ((df['Sex'] == 'M') & (df['Termd'] == 1)).sum()
print(f'Total males terminated: {terminated_male_count}')

Total females terminated: 60
Total males terminated: 44

# Generate a pastel color palette.
pastel_colors = generate_pastel_palette(6)

# Set custom colors for 'Female' and 'Male' categories.
female_color = pastel_colors[2]
male_color = pastel_colors[5]

# Count of terminated male and female employees.
terminated_female_count = ((df['Sex'] == 'F') & (df['Termd'] == 1)).sum()
terminated_male_count = ((df['Sex'] == 'M') & (df['Termd'] == 1)).sum()

# Total terminations.
total_terminations = df['Termd'].sum()

# Calculate the termination rate for male and female employees.
termination_rate_female = terminated_female_count / total_terminations * 100
termination_rate_male = terminated_male_count / total_terminations * 100

# Gender data.
gender_counts = df['Sex'].value_counts()
sizes = gender_counts.values

# Get counts of females and males
female_count = sizes[0] if 'Female' in labels else 0
male_count = sizes[1] if 'Male' in labels else 0

# Define labels for the pie chart
labels = ['Female', 'Male']
explode = (0.05, 0.05)

# Create legend labels with additional text.
legend_labels = [f'{label} (Terminated: {count})' for label, count in zip(labels, [terminated_female_count, terminated_male_count])]

# Pie Chart.
fig, ax = plt.subplots(figsize=(6,6))
wedges, texts, autotexts = plt.pie([termination_rate_female, termination_rate_male], labels=labels, 
                                   colors=[female_color, male_color], 
                                   autopct='%1.2f%%', 
                                   pctdistance=0.80, 
                                   explode=explode,
                                   labeldistance=0.35, 
                                   wedgeprops={'linewidth': 0.3, 'edgecolor': pie_wedge_color},
                                   textprops={'color': annotation_color, 'weight': 'normal', 'fontsize': '12'},
                                   shadow=True)

# Draw inner circle for donut effect.
centre_circle = plt.Circle((0, 0), 0.65, fc='white')
fig = plt.gcf()

# Adding circle in pie chart.
fig.gca().add_artist(centre_circle)

# Adding title of the chart.
plt.title('Gender Attrition Rate by Total Terminations', **title_label_params)

# Add Legends.
plt.legend(legend_labels, loc="upper right", bbox_to_anchor=(1.26, 0.95))

# Set font color of text labels.
for text in texts:
    text.set_color('black')
    text.set_fontsize('12')

# Display the chart.
plt.show()

gender_attrition_df = df.groupby('Sex')['Termd'].mean()
gender_attrition_df.reset_index()

# Generate a pastel color palette.
pastel_colors = generate_pastel_palette(6)

# Set custom colors for 'Female' and 'Male' categories.
female_color = pastel_colors[2]
male_color = pastel_colors[5]

bar_width = 0.5 # custom bar width.

# Plot the bar plot.
fig, ax = plt.subplots(figsize=(6,4))
ax.set_facecolor(chart_bg_color)

# Set y-label ticks to display percentages
yticks = np.arange(0, 1.01, 0.10)
ax.set_yticks(yticks)
ax.set_yticklabels([f'{tick:.0%}' for tick in yticks])

# Get the current tick positions and labels.
ticks = ax.get_xticklabels()

# Adjust the position of the tick labels slightly to the left.
for tick in ticks:
    tick.set_horizontalalignment('center')
    tick.set_x(-0.5)

# Hide spines
ax.spines['top'].set_visible(False)
ax.spines['right'].set_visible(False)

# Plot the bar plot
gender_attrition_df.plot(kind='bar', 
        color=[female_color, male_color], edgecolor=bar_edge_color, linewidth=0.5, width=bar_width)

# Add x-axis labels.
plt.xticks(range(len(gender_attrition_df)), ['Female', 'Male'], rotation=0)

# Annotate bars with percentages.
for i, rate in enumerate(gender_attrition_df.values):
    if rate > 0.00:
        ax.text(i, rate + 0.01, f'{rate:.2%}', ha='center', color='darkblue', fontsize=10)

# Add labels and title
plt.xlabel('', fontsize=12)
plt.ylabel('Attrition Rate', fontweight='bold', labelpad=10)
plt.title('Gender-based Attrition Rates', **title_label_params)

# Display plot
plt.tight_layout()
plt.ylim(0, 0.40)
plt.grid(True, alpha=0.70, axis='y', linestyle=':')
plt.show()

df.groupby(['EmpSatisfaction', 'Termd']).size().reset_index(name='count')

# Generate a pastel color palette.
pastel_colors = generate_pastel_palette(2)

bar_width = 0.90 # custom bar width.

# Plot the bar plot.
fig, ax = plt.subplots(figsize=(8,6))
ax.set_facecolor(chart_bg_color)

# Define labels for the pie chart
labels = ['Remain Employed', 'Terminated']

# Create legend labels with additional text
legend_labels = [f'{label}' for label in labels]

# Create custom legend handles using empty proxy artists.
handles = [plt.Rectangle((0,0),1,1, color=color) for color in pastel_colors]

# Hide spines
ax.spines['top'].set_visible(False)
ax.spines['right'].set_visible(False)
        
# Calculate percentages for each 'Termd' group within each 'EmpSatisfaction' category
percentage_df = df.groupby(['EmpSatisfaction', 'Termd']).size().reset_index(name='count')
percentage_df['percentage'] = percentage_df.groupby('EmpSatisfaction')['count'].transform(lambda x: x / x.sum() * 100)

# Create a grouped bar plot for 'EmpSatisfaction' with 'Termd' as hue
sns.barplot(x='EmpSatisfaction', y='percentage', hue='Termd', data=percentage_df, palette=[pastel_colors[0], pastel_colors[1]],
            width=bar_width, edgecolor=bar_edge_color, linewidth=0.5)

# Annotate each bar with percentages
for p in ax.patches:
    height = p.get_height()
    ax.text(p.get_x() + p.get_width() / 2., height + 0.5, f'{height:.0f}%', ha='center', va='bottom')

# Convert y-axis tick labels to percentages
ticks = ax.get_yticks()
ax.set_yticklabels([f'{tick:.0f}%' for tick in ticks])

# Set the title and labels
plt.title('Employee Satisfaction \nby Termination Status', fontweight='bold', pad=10)
plt.xlabel('Employee Satisfaction Scores', fontweight='bold', labelpad=10)
plt.ylabel('')

# Show the plot
legend = plt.legend(title='Termination Status', labels=legend_labels, handles=handles, loc="upper right", bbox_to_anchor=(0.99, 0.98))
legend.get_title().set_fontweight('bold')
plt.tight_layout()
plt.ylim(0, 110)
plt.grid(True, alpha=0.60, axis='y', linestyle=':')
plt.show()

pearson_coef, p_value = stats.pearsonr(df['EmpSatisfaction'], df['Termd'])
print("The Pearson Correlation Coefficient is", pearson_coef, " with a P-value of P =", p_value)

The Pearson Correlation Coefficient is -0.004731666973705058  with a P-value of P = 0.9337654152178527

df['MaritalDesc'].value_counts()

MaritalDesc
Single       137
Married      124
Divorced      30
Separated     12
Widowed        8
Name: count, dtype: int64

marital_df = df.groupby('MaritalDesc')['Termd'].mean().reset_index(name='AttritionRate').sort_values(by='AttritionRate', ascending=False)
marital_df

# Generate a pastel color palette.
pastel_colors = generate_pastel_palette(len(marital_df))

bar_width = 0.9 # custom bar width.

# Plot the bar plot.
fig, ax = plt.subplots(figsize=(6,4))
ax.set_facecolor(chart_bg_color)

# Set y-label ticks to display percentages
yticks = np.arange(0, 1.01, 0.10)
ax.set_yticks(yticks)
ax.set_yticklabels([f'{tick:.0%}' for tick in yticks])

# Get the current tick positions and labels.
ticks = ax.get_xticklabels()

# Adjust the position of the tick labels slightly to the left.
for tick in ticks:
    tick.set_horizontalalignment('right')
    tick.set_x(-0.5)

# Hide spines
ax.spines['top'].set_visible(False)
ax.spines['right'].set_visible(False)

# Plot the bar plot
plt.bar(marital_df['MaritalDesc'], marital_df['AttritionRate'], 
        color=pastel_colors, edgecolor=bar_edge_color, linewidth=0.5, width=bar_width)

# Annotate bars with percentages
for i, rate in enumerate(marital_df['AttritionRate']):
    if rate > 0.00:
        ax.text(i, rate + 0.01, f'{rate:.0%}', ha='center', color='darkblue', fontsize=9)

# Add labels and title
plt.xlabel('', fontsize=12)
plt.ylabel('Attrition Rate', fontweight='bold', labelpad=10)
plt.title('Marital Status Attrition Rates', **title_label_params)
plt.xticks(rotation=45)

# Display plot
plt.tight_layout()
plt.ylim(0, 0.60)
plt.grid(True, alpha=0.70, axis='y', linestyle=':')
plt.show()

source_counts = df['RecruitmentSource'].value_counts().sort_values(ascending=True)
source_counts.to_frame().reset_index()

# Generate a pastel color palette based on the number of recruitment sources.
pastel_colors = generate_pastel_palette(len(source_counts))

# Define the width of the bars.
bar_width = 0.75

# Plot the bar plot
fig, ax = plt.subplots(figsize=(8,6))
source_counts.plot(kind='bar', color=pastel_colors, edgecolor=bar_edge_color, linewidth=0.5, width=bar_width)
ax.set_facecolor(chart_bg_color)

# Adjust the limits of the y-axis.
ax.set_ylim(0,)

# Hide spines.
ax.spines['top'].set_visible(False) 
ax.spines['right'].set_visible(False)

# Annotate bars with figure just below the top of each bar.
for i, count in enumerate(source_counts):
    ax.text(i, count + 0.2, str(count), ha='center', va='bottom', color=annotation_color)

plt.xlabel('')
plt.ylabel('Count of Sources', fontweight='bold')
plt.title('Recruitment Sources', **title_label_params)
plt.xticks(rotation=45, ha='right')
plt.tight_layout()
plt.grid(True, alpha=0.15)
plt.show()

source_by_race = df.groupby('RaceDesc')['RecruitmentSource'].value_counts().reset_index()
source_by_race

# Create a pivot table with the count of each recruitment source for each race.
pivot_table = source_by_race.pivot_table(index='RaceDesc', columns='RecruitmentSource', values='count', fill_value=0)
pivot_table

# Calculate the proportion of hires from each recruitment source within each race category.
proportion_pivot_table = pivot_table.div(pivot_table.sum(axis=1), axis=0)

# Generate a pastel color palette based on the number of recruitment sources.
pastel_colors = generate_pastel_palette(len(source_counts))

# Plot the bar plot
fig, ax = plt.subplots(figsize=(8,6))
proportion_pivot_table.plot(kind='bar', color=pastel_colors, edgecolor=bar_edge_color, linewidth=0.5, width=bar_width, stacked=True, ax=ax)
ax.set_facecolor(chart_bg_color)

# Plotting.
plt.title('Proportion of Hires by\nRecruitment Source and Race', **title_label_params)
plt.xlabel('')
plt.ylabel('Proportion of Hires', fontweight='bold', labelpad=10)
plt.xticks(rotation=50, ha='right')
plt.legend(title='Recruitment Source', bbox_to_anchor=(1.04, 1), loc='upper left')
plt.tight_layout()
plt.show()

# Generate a pastel color palette based on the number of recruitment sources.
pastel_colors = generate_pastel_palette(df['FromDiversityJobFairID'].nunique())

# Data
labels = ['Not from Job Fair', 'From Job Fair']
sizes = df['FromDiversityJobFairID'].value_counts()

explode = (0.1, 0)  # Explode the first slice.

# Pie Chart.
fig, ax = plt.subplots(figsize=(6,4))
wedges, texts, autotexts = plt.pie(sizes, 
                                   colors=pastel_colors, 
                                   labels=labels, 
                                   autopct='%1.1f%%', 
                                   pctdistance=0.60, 
                                   explode=explode,
                                   labeldistance=1.08, 
                                   startangle=140,
                                   wedgeprops={'linewidth': 0.3, 'edgecolor': pie_wedge_color},
                                   textprops={'color': annotation_color, 'weight': 'normal', 'fontsize': '10'},
                                   shadow=True)

legend =plt.legend(title='', bbox_to_anchor=(1.08, 1), loc='upper right', labels=labels)
legend.get_title().set_fontweight('bold')  # Make legend title bold

# Set font color of text labels.
for text in texts:
    text.set_color('black')
    text.set_fontsize('9')

plt.title('Distribution of Employees Recruited from\nDiversity Job Fair', **title_label_params)
plt.axis('equal')  # Equal aspect ratio ensures that pie is drawn as a circle
plt.tight_layout()
plt.show()

# Copy dataframe for Performance scores manipulation.
performance_df = df.copy()

# Define the mapping dictionary.
score_mapping = {'Exceeds': 3, 'Fully Meets': 2, 'Needs Improvement': 1, 'PIP': 0}

# Replace categorical values with numerical values
performance_df['PerformanceScore'] = performance_df['PerformanceScore'].replace(score_mapping).astype(int)

performance_df['PerformanceScore'].value_counts()

PerformanceScore
2    243
3     37
1     18
0     13
Name: count, dtype: int64

# Average performance score grouped by department.
avg_perf_by_dept = performance_df.groupby('Department')['PerformanceScore'].mean().reset_index()

# Sort by average performance score in ascending order
avg_perf_by_dept = avg_perf_by_dept.sort_values(by='PerformanceScore', ascending=True)

# Generate a pastel color palette based on the number of departments.
num_of_depts = df['Department'].nunique()
pastel_colors = generate_pastel_palette(num_of_depts)

# Define the width of the bars.
bar_width = 0.70

fig, ax = plt.subplots(figsize=(8,6))
sns.barplot(x='Department', y='PerformanceScore', data=avg_perf_by_dept, 
            palette=pastel_colors, 
            alpha=0.85, 
            ax=ax, 
            edgecolor=bar_edge_color, 
            linewidth=0.5, 
            width=bar_width)

ax.set_facecolor(chart_bg_color)

# Hide spines.
ax.spines['top'].set_visible(False) 
ax.spines['right'].set_visible(False)

# Set y-axis limits
ax.set_ylim(0, 3.00)

# Add annotations on top of each bar
for bar in ax.patches:
    height = bar.get_height()
    ax.text(bar.get_x() + bar.get_width() / 2, height, round(height, 2), ha='center', va='bottom', color=annotation_color)

# Create a custom legend
legend_mapping = {'Exceeds': 3, 'Fully Meets': 2, 'Needs Improvement': 1, 'PIP': 0}
legend_labels = [f'{value} - {label}' for label, value in legend_mapping.items()]
legend = ax.legend(labels=legend_labels, title='Score Benchmarks', loc='upper right', bbox_to_anchor=(1.40, 1),handlelength=0)

# Set the legend title font weight to bold
legend.get_title().set_fontweight('bold')

plt.xlabel('')
plt.ylabel('Average Performance Score', fontweight='bold', labelpad=10)
plt.title('Average Performance Score by Department', **title_label_params)
plt.xticks(rotation=45, ha='right')  # Rotate department names for better readability
plt.grid(True, alpha=0.15)
plt.show()

# Calculate mean performance score per manager
manager_performance = performance_df.groupby('ManagerName')['PerformanceScore'].mean().reset_index()

# Sort the dataframe by performance score in ascending order.
manager_performance_sorted = manager_performance.sort_values(by='PerformanceScore', ascending=True)

# Display lowest five scores.
print("Lowest five performance scores:\n")
print(manager_performance_sorted.head(5).to_string(header=False))

# Display highest five scores.
print("\nHighest five performance scores:\n")
print(manager_performance_sorted.tail(5).to_string(header=False))

Lowest five performance scores:

7   Debra Houlihan  1.666667
12      John Smith  1.714286
16   Lynn Daneault  1.846154
18    Peter Monroe  1.857143
17  Michael Albert  1.863636

Highest five performance scores:

10       Janet King  2.105263
0   Alex Sweetwater  2.111111
19       Simon Roup  2.117647
11  Jennifer Zamora  2.142857
9      Eric Dougall  2.250000

# Calculate mean performance score per manager
manager_performance = performance_df.groupby('ManagerName')['PerformanceScore'].mean().reset_index()

# Sort by performance score in ascending order.
manager_performance_sorted = manager_performance.sort_values(by='PerformanceScore', ascending=True)

# Generate a pastel color palette based on the number of managers.
num_of_managers = df['ManagerName'].nunique()
palette = sns.color_palette("RdYlBu", len(manager_performance_sorted)) # Define a color palette

# Define the width of the bars.
bar_width = 0.70

# Calculate the total width of the bars and the gap between them
num_bars = len(ax.patches)
total_width = num_bars * bar_width
total_gap = 1.05
gap_width = (total_gap - total_width) / (num_bars - 1)

# Calculate the x positions of the bars
bar_positions = np.arange(num_bars) * (bar_width + gap_width)

# Plotting
fig, ax = plt.subplots(figsize=(12, 6))
sns.barplot(x='ManagerName', y='PerformanceScore', data=manager_performance_sorted, 
            palette=palette, 
            edgecolor=bar_edge_color, 
            linewidth=0.5,
            width=bar_width)

# Set chart background colour.
ax.set_facecolor(chart_bg_color)

# Add annotations on top of each bar for lowest 5 and top 5
for i, bar in enumerate(ax.patches):
    height = bar.get_height()
    if i < 5 or i >= len(ax.patches) - 5:
        ax.text(bar.get_x() + bar.get_width() / 2, height, round(height, 2), 
                ha='center', va='bottom', color='black', fontsize=10)

# Hide spines.
ax.spines['top'].set_visible(False) 
ax.spines['right'].set_visible(False)

# Create a custom legend with only the lowest 5 scores
lowest_scores = manager_performance_sorted['PerformanceScore'].head(5)
lowest_colors = sns.color_palette(palette, n_colors=5)

# These "dummy" bars are not part of the main plot but serve the purpose of displaying the legend accurately.
for score, color in zip(lowest_scores, lowest_colors):
    plt.bar(0, 0, color=color, label=f'{score:.2f}')
    
# Set legend to upper left.
legend = plt.legend(title='Lowest 5 Scores', loc='upper left', bbox_to_anchor=(0.035, 0.98))

# Set the legend title font weight to bold
legend.get_title().set_fontweight('bold')

# Customize legend background and border
frame = legend.get_frame()
frame.set_facecolor('lightgrey')  # Set background color
frame.set_linewidth(1.1)  # Set border width
frame.set_edgecolor('black')  # Set border color

plt.title('Average Performance Scores Under Each Manager', **title_label_params)
plt.xlabel('')
plt.ylabel('Performance Score', fontweight='bold', labelpad=10)
plt.ylim(0, 3.2)
plt.xticks(rotation=45, ha='right')
plt.tight_layout()
plt.grid(True, alpha=0.15)
plt.show()

engagement_score_sales_mean = df[df['Department'] == 'Sales']['EngagementSurvey'].mean()
print(f'Mean Engagement Score for Sales Department: \t\t{engagement_score_sales_mean:.2f}')

engagement_score_prod_mean = df[df['Department'] == 'Production']['EngagementSurvey'].mean()
print(f'Mean Engagement Score for Production Department: \t{engagement_score_prod_mean:.2f}')

Mean Engagement Score for Sales Department: 		3.82
Mean Engagement Score for Production Department: 	4.13

# Dataframe for 'Sales' department.
sales_df = df[df['Department'] == 'Sales']

# DataFrame for 'Production' department.
production_df = df[df['Department'] == 'Production']

# Generate pastel colors.
pastel_colors = generate_pastel_palette(7)

# Create a figure with two subplots and get axes.
fig, axes = plt.subplots(1, 2, figsize=(15, 6)) #sharex=True, sharey=True) # share same x and y axes scale.

# Add super title.
fig.suptitle("Distribution of Engagement Survey Values", fontweight='bold')

# Plot KDE plot for 'EngagementSurvey' values in the 'Sales' department
sns.kdeplot(data=sales_df['EngagementSurvey'], shade=True, ax=axes[0], color=pastel_colors[3])
axes[0].set_xlabel('Engagement Survey Value', fontweight='bold', labelpad=5)
axes[0].set_ylabel('Density', fontweight='bold', labelpad=5)
axes[0].set_title('Sales Department', color='darkblue')
axes[0].grid(True, alpha=0.15)

# Set background color for the subplot
axes[0].set_facecolor(chart_bg_color)

# Plot KDE plot for 'EngagementSurvey' values in the 'Production' department
sns.kdeplot(data=production_df['EngagementSurvey'], shade=True, ax=axes[1], color=pastel_colors[4])
axes[1].set_xlabel('Engagement Survey Value', fontweight='bold', labelpad=5)
axes[1].set_ylabel('Density', fontweight='bold', labelpad=5)
axes[1].set_title('Production Department', color='darkblue')
axes[1].grid(True, alpha=0.15)

# Set background color for the subplot
axes[1].set_facecolor(chart_bg_color)

# Add vertical lines and annotations to highlight the bimodal distribution
sales_peaks = [2.2, 4.35]  # Example values for peak locations in the Sales department
production_peaks = [4.3]  # Example values for peak locations in the Production department
for ax, peaks, dept_name in zip(axes, [sales_peaks, production_peaks], ['Sales', 'Production']):
    for peak_value in peaks:
        ax.axvline(peak_value, color='red', linestyle='--', alpha=0.7)
        ax.text(peak_value, 0.1, f'Peak ({dept_name})', rotation=90, verticalalignment='bottom', horizontalalignment='right', color='red')

plt.grid(True, alpha=0.15)
plt.show()

sales_managers = df[df['Department'] == 'Sales']['ManagerName'].unique()
sales_managers

array(['John Smith', 'Lynn Daneault', 'Debra Houlihan', 'Janet King'],
      dtype=object)

# Create dataframe with only Sales managers.
sales_manager_df = performance_df[performance_df['Department'] == 'Sales']

# Calculate the average metrics including satisfaction, engagement survey, and performance score for each manager
average_emp_satisfaction = sales_manager_df.groupby('ManagerName')[['EmpSatisfaction', 'EngagementSurvey', 'PerformanceScore']].mean().sort_values(by='EmpSatisfaction')
print(average_emp_satisfaction)

                EmpSatisfaction  EngagementSurvey  PerformanceScore
ManagerName                                                        
John Smith             3.928571          3.786429          1.714286
Janet King             4.000000          4.500000          2.000000
Lynn Daneault          4.076923          3.796154          1.846154
Debra Houlihan         4.333333          3.840000          1.666667

# Group by 'ManagerName' and calculate the mean of 'Termd' to get the attrition rate
attrition_rate = sales_df.groupby('ManagerName')['Termd'].mean().reset_index(name='AttritionRate').sort_values(by='AttritionRate', ascending=False)
attrition_rate

# Count the number of terminated employees for each manager
terminated_count = sales_df[sales_df['Termd'] == 1].groupby('ManagerName')['Termd'].sum().reset_index(name='TermdCount')

# Merge the terminated count with the sales_df DataFrame
sales_df_with_terminated_count = pd.merge(sales_df['ManagerName'].value_counts(), terminated_count, on='ManagerName', how='left')

# Rename the 'count' column to 'total_employees'
sales_df_with_terminated_count = sales_df_with_terminated_count.rename(columns={'count': 'TotalEmployees'})
sales_df_with_terminated_count

# View the statistical summary on Sales Department and Engagement Survey column.
sales_engage_stats = sales_manager_df['EngagementSurvey'].describe()
sales_engage_stats

count    31.000000
mean      3.818710
std       0.952791
min       1.810000
25%       3.270000
50%       4.100000
75%       4.500000
max       5.000000
Name: EngagementSurvey, dtype: float64

# Observe the data for Sales department where Engagement Survey score is within the 25% quartile
Q1_engagement_sales = sales_engage_stats.loc['25%']
sales_q1_engage_df = df[(df['Department'] == 'Sales') & (df['EngagementSurvey'] <= Q1_engagement_sales)]

sales_q1_engage_df[['Employee_Name', 'EmpSatisfaction', 'EngagementSurvey', 'PerformanceScore']]

corr_test = df[['EngagementSurvey','PerfScoreID','EmpID','SpecialProjectsCount', 'ManagerID', 'DaysLateLast30', 'Salary']]
corr_test.corr()

# Remove outliers from Salary in a new dataframe.
df_salary_outliers_removed = df[~potential_outliers_salary].copy() # Outliers removal IQR method.

# Create a figure with two subplots side by side.
fig, axes = plt.subplots(1, 2, figsize=(10, 6)) # 1 row 2 cols

# Generate a pastel color palette with 6 colors to choose 2 from.
pastel_colors = generate_pastel_palette(6)

# Marker data point size
marker_size = 40

# Add super title.
fig.suptitle("Salary vs EngagementSurvey", fontsize=14, fontweight='bold')

# Plot the regression plot for "DaysLateLast30" vs "EngagementSurvey" on the first subplot.
sns.regplot(x="Salary", y="EngagementSurvey", data=df_salary_outliers_removed, 
            ax=axes[0], 
            scatter_kws={'s': marker_size, 'edgecolor': scatter_edge_color},
            line_kws={'color': 'red'})

axes[0].set_title('With Outliers removed')
axes[0].set_ylim(0,)
axes[0].set_facecolor(chart_bg_color)
axes[0].set_xlabel('Salary', fontweight='bold')
axes[0].set_ylabel('EngagementSurvey', fontweight='bold')
axes[0].grid(True, alpha=0.15)

# Plot the regression plot for "Salary" vs "EngagementSurvey" on the second subplot.
sns.regplot(x="Salary", y="EngagementSurvey", data=df, 
            ax=axes[1], 
            scatter_kws={'s': marker_size, 'edgecolor': scatter_edge_color},
            line_kws={'color': 'red'})

axes[1].set_title('With Outliers present')
axes[1].set_ylim(0,)
axes[1].set_facecolor(chart_bg_color)
axes[1].set_xlabel('Salary', fontweight='bold')
axes[1].set_ylabel('EngagementSurvey', fontweight='bold')
axes[1].grid(True, alpha=0.5)

# Adjust layout.
plt.tight_layout()

# Show the plot.
plt.grid(True, alpha=0.15)
plt.show()

pearson_coef, p_value = stats.pearsonr(df['Salary'], df['EngagementSurvey'])
print("The Pearson Correlation Coefficient is", pearson_coef, " with a P-value of P =", p_value)

The Pearson Correlation Coefficient is 0.06496607104732482  with a P-value of P = 0.25333647584714747

# Calculate the mean EngagementSurvey scores for each category of DaysLateLast30
mean_engagement_by_dayslate = df.groupby('DaysLateLast30')['EngagementSurvey'].mean().reset_index()

# Unique count of DaysLateLast30.
dayslate_nunique = df['DaysLateLast30'].nunique()

# Generate a pastel color palette based on the number of termination reasons.
pastel_colors = generate_pastel_palette(dayslate_nunique)

# Plot a bar plot with mean EngagementSurvey scores for each category.
ax = sns.barplot(x='DaysLateLast30', y='EngagementSurvey', data=mean_engagement_by_dayslate, 
                 palette=pastel_colors, 
                 edgecolor=bar_edge_color, 
                 linewidth=0.5)

# Set background color of chart.
ax.set_facecolor(chart_bg_color)

ax.set_xlim(-1.0, 7)
ax.set_ylim(0, df['EngagementSurvey'].max())

# Annotate each bar with the mean value.
for bar in ax.patches:
    ax.annotate(f'{bar.get_height():.2f}', 
                (bar.get_x() + bar.get_width() / 2, bar.get_height()), 
                ha='center', va='bottom', color=annotation_color)

# Hide spines.
ax.spines['top'].set_visible(False) 
ax.spines['right'].set_visible(False)
    
plt.xlabel('Days Late Last 30 Days', fontweight='bold', labelpad=10)
plt.ylabel('Mean Engagement Survey Score', fontweight='bold', labelpad=10)
plt.title('Mean Engagement Survey Score by\nDays Late Last 30 Days', **title_label_params)
plt.xticks(rotation=0)
plt.grid(True, alpha=0.15)
plt.show()

pearson_coef, p_value = stats.pearsonr(df['DaysLateLast30'], df['EngagementSurvey'])
print("The Pearson Correlation Coefficient is", pearson_coef, " with a P-value of P =", p_value)

The Pearson Correlation Coefficient is -0.5852315472719372  with a P-value of P = 5.648682231444894e-30

df['PerformanceScore'].value_counts()

PerformanceScore
Fully Meets          243
Exceeds               37
Needs Improvement     18
PIP                   13
Name: count, dtype: int64

# Group by PerformanceScore and calculate mean salary.
mean_salary_by_perf = df_salary_outliers_removed.groupby('PerformanceScore')['Salary'].mean()
mean_salary_by_perf

PerformanceScore
Exceeds              65046.121212
Fully Meets          65506.829787
Needs Improvement    63196.235294
PIP                  58971.076923
Name: Salary, dtype: float64

# Marker data point size.
marker_size = 20

# Create a figure with two subplots side by side.
fig, axes = plt.subplots(1, 2, figsize=(14,6))

# Add super title.
fig.suptitle("Distribution of Salary by Performance Score", fontweight='bold')

# Iterate over each subplot
for i, ax in enumerate(axes):    
    
    ax.set_facecolor(chart_bg_color)
    ax.grid(True, alpha=0.3)
    ax.set_xlabel('Performance Score', fontweight='bold', labelpad=15)
    ax.set_ylabel('Salary', fontweight='bold', labelpad=10)
    
    # Set title based on subplot index
    title = 'With Outliers removed' if i == 0 else 'With Outliers present'
    ax.set_title(title)
    
    # Choose dataset based on subplot index.
    data = df if i == 1 else df_salary_outliers_removed    
    
    # Plotting the swarm plot
    swarm = sns.swarmplot(x='PerformanceScore', y='Salary', data=data,
                          palette='Set1',
                          s=3,
                          ax=ax)    

# Adjust spacing between subplots
fig.subplots_adjust(wspace=10, hspace=10)

plt.tight_layout()
plt.xlabel('Performance Score')
plt.show()

df.groupby('PerformanceScore')['Salary'].mean()

PerformanceScore
Exceeds              77144.864865
Fully Meets          68366.720165
Needs Improvement    68407.555556
PIP                  58971.076923
Name: Salary, dtype: float64

# Map Performance scores categories to 0 to 3 
perf_df = df.copy()

# Define the mapping dictionary
score_mapping = {'Exceeds': 3, 'Fully Meets': 2, 'Needs Improvement': 1, 'PIP': 0}

# Replace categorical values with numerical values
perf_df['PerformanceScore'] = perf_df['PerformanceScore'].replace(score_mapping).astype(int)

perf_df['PerformanceScore'].value_counts()

PerformanceScore
2    243
3     37
1     18
0     13
Name: count, dtype: int64

pearson_coef, p_value = stats.pearsonr(perf_df['Salary'], perf_df['PerformanceScore'])
print("The Pearson Correlation Coefficient is", pearson_coef, " with a P-value of P =", p_value)

The Pearson Correlation Coefficient is 0.12513801165277122  with a P-value of P = 0.027340106160422525

salary_mean = df['Salary'].mean()
print(f'Mean salary is ${salary_mean.round(2)}')

Mean salary is $69020.68

# Calculate the mean employee satisfaction level for those above and below the average salary.
salary_mean = df['Salary'].mean()
emp_satisfaction_above_mean = df[df['Salary'] >= salary_mean]['EmpSatisfaction'].mean().round(2)
emp_satisfaction_below_mean = df[df['Salary'] < salary_mean]['EmpSatisfaction'].mean().round(2)

print('Mean Employee Satisfaction Level:\n' + '*'*33)
print(f'Above the Salary Average: \t{emp_satisfaction_above_mean}')
print(f'Below the Salary Average: \t{emp_satisfaction_below_mean}\n')

# Calculate lower quartile 25% and upper quartile 75% for Salary.
salary_description = df['Salary'].describe()
Q1_salary = df['Salary'].quantile(0.25)
Q3_salary = df['Salary'].quantile(0.75)

# Select salaries within Q1 in the dataframe.
below_Q1_df = df[df['Salary'] <= Q1_salary]
emp_satisfaction_within_Q1_mean = below_Q1_df['EmpSatisfaction'].mean().round(2)

# Filter the dataframe for salaries within Q3.
below_Q3_df = df[df['Salary'] <= Q3_salary]
emp_satisfaction_within_Q3_mean = below_Q3_df['EmpSatisfaction'].mean().round(2)

print(f'Within the 25th percentile: \t{emp_satisfaction_within_Q1_mean}')
print(f'Within the 75th percentile: \t{emp_satisfaction_within_Q3_mean}')

Mean Employee Satisfaction Level:
*********************************
Above the Salary Average: 	3.98
Below the Salary Average: 	3.86

Within the 25th percentile: 	3.85
Within the 75th percentile: 	3.88

# Calculate the mean Salary for each Employment Satisfaction score.
salary_by_empsatisfaction = df.groupby('EmpSatisfaction')['Salary'].mean().reset_index()

# Unique count of DaysLateLast30.
emp_sat_nunique = df['EmpSatisfaction'].nunique()

# Generate a pastel color palette based on the number of satisfaction scores.
pastel_colors = generate_pastel_palette(emp_sat_nunique)

# Plot a bar plot with mean salary scores for each category.
ax = sns.barplot(x='EmpSatisfaction', y='Salary', data=salary_by_empsatisfaction, 
                 palette=pastel_colors,
                 edgecolor=bar_edge_color, 
                 linewidth=0.5)

# Set background color of chart.
ax.set_facecolor(chart_bg_color)

# Annotate each bar with the mean value.
for bar in ax.patches:
    ax.annotate(f'{bar.get_height():,.0f}', 
               (bar.get_x() + bar.get_width() / 2, bar.get_height()), 
                ha='center', va='bottom', color=annotation_color)

# Hide spines.
ax.spines['top'].set_visible(False) 
ax.spines['right'].set_visible(False)
    
plt.xlabel('Employee Satisfaction Score', fontweight='bold', labelpad=10)
plt.ylabel('Mean Salary', fontweight='bold', labelpad=10)
plt.title('Mean Salary by Employee Satisfaction', **title_label_params)
plt.xticks(rotation=0)
plt.grid(True, alpha=0.15)
plt.show()

# Define the interquartile range (IQR).
IQR = Q3_salary - Q1_salary

# Define the upper bound threshold
threshold = 1.5

# Calculate the upper bound
upper_bound = Q3_salary + (threshold * IQR)

# Filter the dataframe for salaries within the upper whisker.
filtered_df = df[(df['Salary'] >= Q3_salary) & (df['Salary'] <= upper_bound)]

# Calculate the mean Employee Satisfaction Level within the filtered subset
emp_satisfaction_within_whisker_mean = filtered_df['EmpSatisfaction'].mean()
print(f'Mean Employee Satisfaction Level within the upper whisker: {emp_satisfaction_within_whisker_mean}')

Mean Employee Satisfaction Level within the upper whisker: 3.795918367346939

# Define the bins for salary.
bins = np.arange(40000, df['Salary'].max() + 10000, 10000)

grouped_by_salary_bins = pd.cut(df['Salary'], bins)

# Create salary bins and calculate mean satisfaction rate for each bin.
satisfaction_by_salary_bins = df.groupby(grouped_by_salary_bins, observed=True)['EmpSatisfaction'].mean().reset_index(name='Satisfaction_Mean') # observed=True Suppress Warning on NaN values.
print(satisfaction_by_salary_bins)

              Salary  Satisfaction_Mean
0     (40000, 50000]           4.032258
1     (50000, 60000]           3.755556
2     (60000, 70000]           3.900990
3     (70000, 80000]           3.967742
4     (80000, 90000]           3.529412
5    (90000, 100000]           4.187500
6   (100000, 110000]           4.181818
7   (110000, 120000]           3.750000
8   (130000, 140000]           5.000000
9   (140000, 150000]           4.500000
10  (150000, 160000]           3.000000
11  (170000, 180000]           4.666667
12  (220000, 230000]           5.000000
13  (240000, 250000]           3.000000

salary_bin_count = len(satisfaction_by_salary_bins)

# Maximum mean satisfaction value for salary bins.
max_satisfaction_rate = satisfaction_by_salary_bins['Satisfaction_Mean'].max()

# Generate a pastel color palette based on the number of salary bins.
pastel_colors = generate_pastel_palette(salary_bin_count)

# Plotting
fig, ax = plt.subplots(figsize=(10, 6))
ax.bar(range(salary_bin_count), satisfaction_by_salary_bins['Satisfaction_Mean'], 
        align='center', 
        color=pastel_colors,
        edgecolor=bar_edge_color, 
        linewidth=0.5)

# Set yticks and manually format ytick labels
ax.set_yticks(range(1, int(max_satisfaction_rate) + 1))  # Set yticks to integers from 1 to the maximum satisfaction rate
ax.set_yticklabels(['{:.1f}'.format(y) for y in range(1, int(max_satisfaction_rate) + 1)])  # Format tick labels with one decimal place

# Get the current tick positions and labels
ticks = ax.get_xticklabels()

# Adjust the position of the tick labels slightly to the left.
for tick in ticks:
    tick.set_horizontalalignment('right')
    tick.set_x(-0.5)

# Annotate each bar with the mean value.
for bar in ax.patches:
    ax.annotate(f'{bar.get_height():.2f}', 
                (bar.get_x() + bar.get_width() / 2, bar.get_height()), 
                ha='center', va='bottom', color=annotation_color)

# Set background color of chart.
ax.set_facecolor(chart_bg_color)

# Hide spines.
ax.spines['top'].set_visible(False) 
ax.spines['right'].set_visible(False)

plt.xlabel('Salary Bins', fontweight='bold', labelpad=10)
plt.ylabel('Mean Satisfaction Score', fontweight='bold', labelpad=10)
plt.title('Mean Satisfaction Score\nby Salary Bins', **title_label_params)

# Format the x-axis ticks to have 1000 comma separators and zero decimal points.
plt.xticks(range(len(satisfaction_by_salary_bins)), ['${:,.0f}-{:,.0f}'.format(int(b.left), int(b.right)) for b in satisfaction_by_salary_bins['Salary']], rotation=45)

# Adjust y-axis limit to extend past maximum rate.
ax.set_ylim(0, max_satisfaction_rate + 0.5)

plt.tight_layout()
plt.grid(True, alpha=0.15)
plt.show()

# Employees where the salary is within the specified range.
positions_in_salary_range = df[(df['Salary'] >= 150000) & (df['Salary'] < 160000)][['Position', 'Department', 'ManagerName', \
                                                                                    'EmpSatisfaction', 'PerformanceScore', 'Salary']]
positions_in_salary_range

# Employees where the salary is within the high specified range.
positions_in_salary_range = df[df['Salary'] >= 240000][['Position', 'EmpSatisfaction', 'Salary']]
positions_in_salary_range

# Employees where the salary is within the medium specified range and satisfaction level of 3.
positions_in_salary_range = df[(df['Salary'] >= 80000) & (df['Salary'] < 90000) & (df['EmpSatisfaction'] == 3)][['Position', 'Department', \
                                                                                                                 'ManagerName', 'EmpSatisfaction', 'PerformanceScore', 
                                                                                                                 'Absences', 'Sex', 'Salary']]

# Display the positions
positions_in_salary_range

# Select workers who are terminated and with employee satisfaction score of 5. 
employed_df = df[(df['TermReason'] != 'N/A-StillEmployed') & (df['EmpSatisfaction'] == 5)]

# Calculate percentages for each 'EmpSatisfaction' group within each Termination Reason category.
percentage_df = employed_df.groupby(['TermReason', 'EmpSatisfaction']).size().reset_index(name='count')
percentage_df['percentage'] = percentage_df.groupby('EmpSatisfaction')['count'].transform(lambda x: x / x.sum() * 100)

# Generate a pastel color palette based on the number of termination reasons.
pastel_colors = generate_pastel_palette(len(percentage_df['TermReason']))

# Create a grouped bar plot for Termination Reasons with 'EmpSatisfaction' as hue.
fig, ax = plt.subplots(figsize=(6,4))
sns.barplot(x='EmpSatisfaction', y='percentage', hue='TermReason', data=percentage_df, 
            # orient='h', 
            palette=pastel_colors,
            edgecolor=bar_edge_color, 
            linewidth=0.5)

# Set background color of chart.
ax.set_facecolor(chart_bg_color)

# Set the title and labels
plt.title('Termination Reasons by\nEmployee Satisfaction Score=5', fontweight='bold', pad=15)
#plt.xlabel('EmpSatisfaction')
ax.set_xticks([])
ax.set_xticklabels([])
ax.set_xlabel('') # Remove x-axis ticks and labels
plt.ylabel('Percentage', fontweight='bold', labelpad=10)

# Show the plot
plt.legend(title='Reasons', loc='center left', bbox_to_anchor=(1, 0.5))
plt.tight_layout()
plt.grid(True, alpha=0.15)
plt.show()

# Capitalise the first letter of each label.
percentage_df['TermReason'] = percentage_df['TermReason'].apply(lambda x: x.capitalize())

# Generate a pastel color palette based on the number of termination reasons.
pastel_colors = generate_pastel_palette(len(percentage_df['TermReason']))

# Function to wrap labels if percentage is less than 4.
def wrap_label(label, percentage, max_percentage=4):
    if percentage < max_percentage:
        words = label.split()  # Split label into words
        wrapped_label = '\n'.join(words)  # Join words with newline characters
        return wrapped_label
    else:
        return label

# Apply the function to the labels
labels = [wrap_label(label, percentage) for label, percentage in zip(percentage_df['TermReason'], percentage_df['percentage'])]

# Plotting the treemap
plt.figure(figsize=(12, 8))
ax = squarify.plot(sizes=percentage_df['percentage'], label=labels, 
              alpha=0.8,
              text_kwargs={'wrap': True, 'fontsize': 11},
              color=pastel_colors,
              edgecolor='darkblue',  # Specify the edge color
              linewidth=2) # Specify the width of the border)
        
plt.axis('off')  # Turn off axis
plt.title('Termination Reasons for Employee Satisfaction Score=5', **title_label_params)
plt.show()

# Generate a pastel color palette with 6 colors to choose 2 from.
pastel_colors = generate_pastel_palette(6)

# Set the background color of the chart
plt.figure(figsize=(10,6))

# Create the box plot for combined salaries
ax = sns.boxplot(x='Department', y='Salary', data=df, 
            palette=pastel_colors)

ax.set_facecolor(chart_bg_color)

# Format the tick values y as a string with dollar sign, comma separator, and zero decimal places.
yticks = np.arange(0, max_ytick + 1, 25000)
ytick_labels = ['${:,.0f}'.format(y) for y in yticks]

# Set yticks and yticklabels
ax.set_yticks(yticks)
ax.set_yticklabels(ytick_labels)

# Set the title and labels
plt.title('Salary Distribution Across Departments', fontweight='bold', pad=10)
plt.ylabel('Salary', fontweight='bold', labelpad=10)
plt.xlabel('')
plt.tight_layout()
plt.grid(True,alpha=0.25)

# Show the plot
plt.show()

department_summary = df.groupby('Department')['Salary'].describe()
department_summary

pearson_coef, p_value = stats.pearsonr(perf_df['Absences'], perf_df['SpecialProjectsCount'])
print("The Pearson Correlation Coefficient is", pearson_coef, " with a P-value of P =", p_value)

The Pearson Correlation Coefficient is -0.020452484246077974  with a P-value of P = 0.7193940929309436

absent_by_project_count = df.groupby('SpecialProjectsCount')['Absences'].mean().reset_index()
absent_by_project_count

# Generate a pastel color palette.
pastel_colors = generate_pastel_palette(2)

# Create line area chart
plt.figure(figsize=(10, 6))
ax = sns.lineplot(x='SpecialProjectsCount', y='Absences', data=absent_by_project_count, 
             marker='o',
             markersize=10,
             color=pastel_colors[1], 
             linewidth=3)

ax.set_facecolor(chart_bg_color)
plt.fill_between(absent_by_project_count['SpecialProjectsCount'], absent_by_project_count['Absences'], color=pastel_colors[1], alpha=0.3)

# Add annotations on markers.
for index, row in absent_by_project_count.iterrows():
    ax.annotate(f"{row['Absences']:.2f}", (row['SpecialProjectsCount'], row['Absences']), 
                textcoords="offset points", xytext=(0,10), ha='center', color=annotation_color)
        
# Add labels and title
plt.title('Average Absences by Special Projects Count', fontsize=16, fontweight='bold', pad=15)
plt.xlabel('Special Projects Count', labelpad=10, fontweight='bold')
plt.ylabel('Average Absences', labelpad=10, fontweight='bold')

# Add padding after the maximum marker location
max_absences = absent_by_project_count['Absences'].max()
plt.ylim(0, max_absences * 1.1)  # Increase the y-axis limit by 10%

# Show plot
plt.grid(True, alpha=0.3)
plt.tight_layout()
plt.show()

projects_df = df.copy()

# Define the mapping dictionary
score_mapping = {'Exceeds': 3, 'Fully Meets': 2, 'Needs Improvement': 1, 'PIP': 0}

# Replace categorical values with numerical values
projects_df['PerformanceScore'] = projects_df['PerformanceScore'].replace(score_mapping).astype(int)
projects_df['PerformanceScore'].value_counts()

PerformanceScore
2    243
3     37
1     18
0     13
Name: count, dtype: int64

perfscore_by_project_count = projects_df.groupby('SpecialProjectsCount')['PerformanceScore'].mean()
perfscore_by_project_count

SpecialProjectsCount
0    1.962656
1    0.000000
2    2.000000
3    2.333333
4    1.888889
5    2.238095
6    1.904762
7    2.083333
8    2.000000
Name: PerformanceScore, dtype: float64

# Group by SpecialProjectsCount and COUNT the number of performance scores.
performance_data = df.groupby(['SpecialProjectsCount', 'PerformanceScore']).size().reset_index(name='Count')
performance_data

# Group by 'SpecialProjectsCount' and calculate the SUM of counts for each Project count.
performance_data['SumCounts'] = performance_data.groupby('SpecialProjectsCount')['Count'].transform('sum')

# Calculate normalised count for each performance score category within each project count.
performance_data['NormalisedCount'] = performance_data['Count'] / performance_data['SumCounts']
print(performance_data)

    SpecialProjectsCount   PerformanceScore  Count  SumCounts  NormalisedCount
0                      0            Exceeds     29        241         0.120332
1                      0        Fully Meets    185        241         0.767635
2                      0  Needs Improvement     16        241         0.066390
3                      0                PIP     11        241         0.045643
4                      1                PIP      1          1         1.000000
5                      2        Fully Meets      1          1         1.000000
6                      3            Exceeds      1          3         0.333333
7                      3        Fully Meets      2          3         0.666667
8                      4        Fully Meets      8          9         0.888889
9                      4  Needs Improvement      1          9         0.111111
10                     5            Exceeds      5         21         0.238095
11                     5        Fully Meets     16         21         0.761905
12                     6            Exceeds      1         21         0.047619
13                     6        Fully Meets     18         21         0.857143
14                     6  Needs Improvement      1         21         0.047619
15                     6                PIP      1         21         0.047619
16                     7            Exceeds      1         12         0.083333
17                     7        Fully Meets     11         12         0.916667
18                     8        Fully Meets      2          2         1.000000

# Pivot the data to have 'PerformanceScore' as columns
pivot_table = performance_data.pivot(index='SpecialProjectsCount', columns='PerformanceScore', values='NormalisedCount').fillna(0)

# Sort rows by special projects count in ascending order
pivot_table_sorted = pivot_table.sort_index(ascending=False)

# Sort columns by performance score in descending order
pivot_table_sorted = pivot_table_sorted.sort_index(axis=1, ascending=False)

# Plot heatmap
plt.figure(figsize=(10, 6))
sns.heatmap(pivot_table_sorted, cmap='YlGnBu', annot=True, fmt=".2f")
plt.xlabel('Performance Score', labelpad=15, fontweight='bold')
plt.ylabel('Special Projects Count', labelpad=15, fontweight='bold')
plt.title('Normalised Distribution of Performance Score\nfor Special Projects Count', pad=15, fontweight='bold')
plt.show()

	Employee_Name	EmpID	MarriedID	MaritalStatusID	GenderID	EmpStatusID	DeptID	PerfScoreID	Salary	...	EngagementSurvey	EmpSatisfaction	SpecialProjectsCount	LastPerformanceReview_Date	Absences	Parsed_DOB	Year	Month	Day
0	Adinolfi, Wilson K	10026	0	0	1	1	5	4	62506	...	4.60	5	0	1/17/2019	1	1983-07-10	1983	7	10
1	Ait Sidi, Karthikeyan	10084	1	1	1	5	3	3	104437	...	4.96	3	6	2/24/2016	17	1975-05-05	1975	5	5
2	Akinkuolie, Sarah	10196	1	1	0	5	5	3	64955	...	3.02	3	0	5/15/2012	3	1988-09-19	1988	9	19
3	Alagbe,Trina	10088	1	1	0	1	5	3	64991	...	4.84	5	0	1/3/2019	15	1988-09-27	1988	9	27
4	Anderson, Carol	10069	0	2	0	5	5	3	50825	...	5.00	4	0	2/1/2016	2	1989-09-08	1989	9	8

	mean	median
Sex
F	67786.727273	62066.5
M	70629.400000	63353.0

	EngagementSurvey	PerfScoreID	EmpID	SpecialProjectsCount	ManagerID	DaysLateLast30	Salary
EngagementSurvey	1.000000	0.544927	-0.589664	0.013227	-0.003344	-0.585232	0.064966
PerfScoreID	0.544927	1.000000	-0.691348	0.045677	-0.060552	-0.734728	0.130903
EmpID	-0.589664	-0.691348	1.000000	-0.043730	0.090236	0.495513	-0.115319
SpecialProjectsCount	0.013227	0.045677	-0.043730	1.000000	-0.525087	-0.092494	0.508333
ManagerID	-0.003344	-0.060552	0.090236	-0.525087	1.000000	0.059882	-0.435406
DaysLateLast30	-0.585232	-0.734728	0.495513	-0.092494	0.059882	1.000000	-0.069443
Salary	0.064966	0.130903	-0.115319	0.508333	-0.435406	-0.069443	1.000000

	count	mean	std	min	25%	50%	75%	max
Department
Admin Offices	9.0	71791.888889	21706.397164	49920.0	55000.00	63003.0	93046.0	106367.0
Executive Office	1.0	250000.000000	NaN	250000.0	250000.00	250000.0	250000.0	250000.0
IT/IS	50.0	97064.640000	33209.612190	50178.0	77417.75	92328.5	106844.5	220450.0
Production	209.0	59953.545455	11422.656822	45046.0	53018.00	59472.0	64066.0	170500.0
Sales	31.0	69061.258065	21271.393236	55875.0	61561.50	65310.0	70506.5	180000.0
Software Engineering	11.0	94989.454545	9566.505698	77692.0	89601.50	95660.0	100807.5	108987.0

	SpecialProjectsCount	Absences
0	0	10.331950
1	1	20.000000
2	2	3.000000
3	3	6.333333
4	4	11.333333
5	5	8.952381
6	6	9.428571
7	7	11.500000
8	8	13.000000

	Employee_Name	EmpID	MarriedID	MaritalStatusID	GenderID	EmpStatusID	DeptID	PerfScoreID	FromDiversityJobFairID	Salary	...	EmpSatisfaction	LastPerformanceReview_Date	Absences	Parsed_DOB	Year	Month	Day	Age
296	Wallace, Courtney E	10274	1	1	0	5	5	3	1	80512	...	3	1/2/2012	5	1955-11-14	1955	11	14	69
288	Valentin,Jackie	10205	1	1	0	1	6	3	0	57859	...	3	1/17/2019	16	1991-05-23	1991	5	23	33
295	Walker, Roger	10190	0	0	1	1	5	3	0	66541	...	5	2/12/2019	4	1976-02-10	1976	2	10	48

	Employee_Name	Position	State	DOB	Sex	MaritalDesc	CitizenDesc	HispanicLatino	RaceDesc	DateofHire	DateofTermination	TermReason	EmploymentStatus	Department	ManagerName	RecruitmentSource	PerformanceScore	LastPerformanceReview_Date	Parsed_DOB
0	Adinolfi, Wilson K	Production Technician I	MA	07/10/83	M	Single	US Citizen	No	White	7/5/2011	NaN	N/A-StillEmployed	Active	Production	Michael Albert	LinkedIn	Exceeds	1/17/2019	1983-07-10
1	Ait Sidi, Karthikeyan	Sr. DBA	MA	05/05/75	M	Married	US Citizen	No	White	3/30/2015	6/16/2016	career change	Voluntarily Terminated	IT/IS	Simon Roup	Indeed	Fully Meets	2/24/2016	1975-05-05
2	Akinkuolie, Sarah	Production Technician II	MA	09/19/88	F	Married	US Citizen	No	White	7/5/2011	9/24/2012	hours	Voluntarily Terminated	Production	Kissy Sullivan	LinkedIn	Fully Meets	5/15/2012	1988-09-19

	EmpID	MarriedID	MaritalStatusID	GenderID	EmpStatusID	DeptID	PerfScoreID	FromDiversityJobFairID	Salary	Termd	...	ManagerID	EngagementSurvey	EmpSatisfaction	SpecialProjectsCount	DaysLateLast30	Absences	Year	Month	Day	Age
count	311.000000	311.000000	311.000000	311.000000	311.000000	311.000000	311.000000	311.000000	311.000000	311.000000	...	303.000000	311.000000	311.000000	311.000000	311.000000	311.000000	311.000000	311.000000	311.000000	311.000000
mean	10156.000000	0.398714	0.810289	0.434084	2.392283	4.610932	2.977492	0.093248	69020.684887	0.334405	...	14.570957	4.110000	3.890675	1.218650	0.414791	10.237942	1978.591640	6.659164	14.971061	45.408360
std	89.922189	0.490423	0.943239	0.496435	1.794383	1.083487	0.587072	0.291248	25156.636930	0.472542	...	8.078306	0.789938	0.909241	2.349421	1.294519	5.852596	8.870236	3.202757	8.734939	8.870236
min	10001.000000	0.000000	0.000000	0.000000	1.000000	1.000000	1.000000	0.000000	45046.000000	0.000000	...	1.000000	1.120000	1.000000	0.000000	0.000000	1.000000	1951.000000	1.000000	1.000000	32.000000
25%	10078.500000	0.000000	0.000000	0.000000	1.000000	5.000000	3.000000	0.000000	55501.500000	0.000000	...	10.000000	3.690000	3.000000	0.000000	0.000000	5.000000	1973.000000	4.000000	7.000000	38.000000
50%	10156.000000	0.000000	1.000000	0.000000	1.000000	5.000000	3.000000	0.000000	62810.000000	0.000000	...	15.000000	4.280000	4.000000	0.000000	0.000000	10.000000	1980.000000	7.000000	15.000000	44.000000
75%	10233.500000	1.000000	1.000000	1.000000	5.000000	5.000000	3.000000	0.000000	72036.000000	1.000000	...	19.000000	4.700000	5.000000	0.000000	0.000000	15.000000	1986.000000	9.000000	22.000000	51.000000
max	10311.000000	1.000000	4.000000	1.000000	5.000000	6.000000	4.000000	1.000000	250000.000000	1.000000	...	39.000000	5.000000	5.000000	8.000000	6.000000	20.000000	1992.000000	12.000000	31.000000	73.000000

	Employee_Name	Position	State	DOB	Sex	MaritalDesc	CitizenDesc	HispanicLatino	RaceDesc	DateofHire	DateofTermination	TermReason	EmploymentStatus	Department	ManagerName	RecruitmentSource	PerformanceScore	LastPerformanceReview_Date	Parsed_DOB
count	311	311	311	311	311	311	311	311	311	311	104	311	311	311	311	311	311	311	311
unique	311	31	28	307	2	5	3	4	6	101	96	18	3	6	21	9	4	137	307
top	Adinolfi, Wilson K	Production Technician I	MA	09/09/65	F	Single	US Citizen	No	White	1/10/2011	8/19/2018	N/A-StillEmployed	Active	Production	Michael Albert	Indeed	Fully Meets	1/14/2019	1965-09-09
freq	1	137	276	2	176	137	295	282	187	14	2	207	207	209	22	87	243	18	2

	Position	Sex	Salary
0	Accountant I	F	63000.000000
1	Accountant I	M	63761.500000
3	Area Sales Manager	F	63793.500000
4	Area Sales Manager	M	65843.800000
5	BI Developer	F	95920.000000
6	BI Developer	M	95313.333333
9	Data Analyst	F	90909.666667
10	Data Analyst	M	89065.200000
12	Database Administrator	F	107124.750000
13	Database Administrator	M	113999.000000
21	IT Support	F	63188.000000
22	IT Support	M	64511.666667
23	Network Engineer	F	51674.500000
24	Network Engineer	M	68225.333333
27	Production Manager	F	72946.666667
28	Production Manager	M	77055.375000
29	Production Technician I	F	56081.578313
30	Production Technician I	M	54667.425926
31	Production Technician II	F	65397.891892
32	Production Technician II	M	63956.700000
33	Sales Manager	F	67364.000000
34	Sales Manager	M	72992.000000
35	Senior BI Developer	F	81584.000000
36	Senior BI Developer	M	86412.000000
38	Software Engineer	F	96906.166667
39	Software Engineer	M	96438.750000
42	Sr. DBA	F	100031.000000
43	Sr. DBA	M	104437.000000
44	Sr. Network Engineer	F	96127.000000
45	Sr. Network Engineer	M	91033.333333

	Age	AttritionRate
20	52	0.571429
36	72	0.666667
28	60	0.800000
21	53	1.000000
31	65	1.000000
33	69	1.000000
35	71	1.000000

	Department	AttritionRate
0	Admin Offices	22.22
1	Executive Office	0.00
2	IT/IS	20.00
3	Production	39.71
4	Sales	16.13
5	Software Engineering	36.36

	RaceDesc	Termd
0	American Indian or Alaska Native	0.000000
3	Hispanic	0.000000
4	Two or more races	0.272727
1	Asian	0.310345
5	White	0.336898
2	Black or African American	0.362500

	RaceDesc	TermReason	Count
0	Asian	Another position	3
1	Asian	unhappy	3
2	Asian	more money	1
3	Black or African American	Another position	7
4	Black or African American	career change	4
5	Black or African American	attendance	3
6	Two or more races	Another position	1
7	Two or more races	performance	1
8	Two or more races	relocation out of area	1
9	White	Another position	9
10	White	more money	9
11	White	unhappy	9

	ManagerName	AttritionRate
20	Webster Butler	0.619048
1	Amy Dunn	0.619048
15	Kissy Sullivan	0.545455
19	Simon Roup	0.470588
17	Michael Albert	0.409091

	Sex	Termd
0	F	0.340909
1	M	0.325926

	RecruitmentSource	count
0	On-line Web application	1
1	Other	2
2	Website	13
3	CareerBuilder	23
4	Diversity Job Fair	29
5	Employee Referral	31
6	Google Search	49
7	LinkedIn	76
8	Indeed	87

RecruitmentSource	CareerBuilder	Diversity Job Fair	Employee Referral	Google Search	Indeed	LinkedIn	On-line Web application	Other	Website
RaceDesc
American Indian or Alaska Native	0.0	0.0	0.0	1.0	1.0	1.0	0.0	0.0	0.0
Asian	2.0	0.0	1.0	7.0	10.0	8.0	0.0	0.0	1.0
Black or African American	5.0	29.0	5.0	4.0	17.0	16.0	0.0	0.0	4.0
Hispanic	0.0	0.0	0.0	0.0	1.0	0.0	0.0	0.0	0.0
Two or more races	0.0	0.0	0.0	2.0	4.0	4.0	0.0	0.0	1.0
White	16.0	0.0	25.0	35.0	54.0	47.0	1.0	2.0	7.0

	Employee_Name	EmpSatisfaction	EngagementSurvey	PerformanceScore
32	Bunbury, Jessica	5	3.14	Fully Meets
45	Chigurh, Anton	5	3.00	Fully Meets
67	Delarge, Alex	3	1.93	PIP
72	Dietrich, Jenna	1	2.30	PIP
95	Forrest, Alex	5	2.00	PIP
258	Smith, John	4	2.40	Needs Improvement
267	Stansfield, Norman	2	1.81	PIP
288	Valentin,Jackie	3	2.81	Fully Meets

	ManagerName	AttritionRate	EngagementSurvey
20	Webster Butler	0.619048	4.325238
1	Amy Dunn	0.619048	3.924762
15	Kissy Sullivan	0.545455	4.039091
19	Simon Roup	0.470588	4.225294
17	Michael Albert	0.409091	4.066818

	MaritalDesc	AttritionRate
0	Divorced	0.533333
4	Widowed	0.500000
1	Married	0.379032
3	Single	0.262774
2	Separated	0.083333

	ManagerName	AttritionRate
0	Debra Houlihan	0.333333
2	John Smith	0.214286
3	Lynn Daneault	0.076923
1	Janet King	0.000000

	Position	Department	ManagerName	EmpSatisfaction	PerformanceScore	Salary
190	IT Manager - Infra	IT/IS	Jennifer Zamora	3	Needs Improvement	157000
240	Data Architect	IT/IS	Brian Champaigne	3	Fully Meets	150290

Unlocking Insights: Exploring Human Resources Dynamics¶

Table of Contents:¶

Introduction¶

Data Description¶

Project Objectives¶

Data Preparation¶

Initial Pre-processing Phase¶

Integrity - Basic Data Quality Checks¶

Remove trailing spaces in object columns¶

Handling Two-Digit Year (DOB) Ambiguity in CSV Dataset¶

Testing and Confirming parsing dates using custom function¶

Apply parse date function on original dataframe¶

Calculate and Add Age attribute to Dataframe¶

Exploratory Data Analysis (EDA)¶

Overview of the Dataset¶

Descriptive Statistical Analysis¶

Summary Statistics¶

Distributions for Continuous Variables¶

Potential Outliers Detection Using IQR (Interquartile Range) Method¶

General Correlation Analysis¶

Visual EDA¶

Distribution of Employment Status¶

Termination Reasons Distribution¶

Employee Distribution Across Departments¶

Diversity Profile¶

Gender Distribution¶

Race Distribution¶

Age Distribution¶

Overall Performance Score Distribution¶

Gender and Position-Based Salary Analysis¶

Box Violin Charts of Gender Salaries¶

Closer look at the Box Plots for Gender Salaries Distribution¶

Distribution of Employees Specific Salary Ranges by Gender¶

Salary by Position¶

Attrition Rate and Turnover Analysis¶

Attrition by Age¶

Attrition Rate by Age Bins¶

Department Attrition Rate and Termination Reasons¶

Attrition Rate by Department¶

Termination Reasons by Department¶

Group Termination Reasons into Broader Categories¶

Attrition Rate by Race Description¶

Race Attrition Rate impact on Total Workforce¶

Top 3 Termination Reasons by Race¶

Attrition Rate by Special Projects Count¶

Identifying the Peak Year of Terminations¶

Attrition Rate in Years of Service¶

Attrition Rate by Salary Bins¶

Attrition Rate by Manager¶

Top 5 Manager Attrition Rates¶

Manager-wise Attrition Rates¶

Gender-based Attrition Rate¶

Relative to Total Terminations¶

Gender-based Attrition Rates¶

Employee Satisfaction by Termination Status¶

Attrition Rate by Martical Description¶

Recruitment Source Distribution and by Race Analysis¶

Exploring Department and Manager-Level Performance-Analysis¶

Average Performance Score per Department¶

Average Performance Score per Manager¶

Engagement Survey Distribution for Sales and Production¶

Performances by Sales Manager and Sales Employees¶

Extending Correlation Analysis: A Deeper Dive¶

Salary and EngagementSurvey¶

DaysLateLast30 and EngagementSurvey¶

Salary and PerformanceScore analysis¶

Swarm plots¶

Encoding ordinal categorical variable 'PerformanceScore'¶

Salary and Employee Satisfaction analysis¶

Investigating Departure Among Highly Satisfied Employees¶

Salary Distribution across Departments¶

Absences by Special Projects Count¶

Special Projects Count and Performance Score¶

Calculate normalised count for each performance score category within each project count¶

Conclusion¶

Summary of the analysis¶

Interesting Key Findings¶

Actionable Insights¶